전체 AI 논문 - 2026-09-15

1. Corrupt Plans, Clean Traces: Evading Chain-of-Thought Monitoring with Plan Injection

• Authors: Keertana Chidambaram , Andrew Ilyas , Vasilis Syrgkanis
• URL: https://arxiv.org/abs/2609.15989
• Abstract:

  Chain-of-thought (CoT) monitoring is a safety strategy where the reasoning of a large language model “actor” is inspected by a “monitor” (often another language model) for signs of unsafe planning, deception, or misalignment. We find that planting harmful but benign-sounding reasoning in the actor’s context can steer it to perform adversarial actions while evading monitors, an attack we term “plan injection”. We initially discover this attack in the multiple-choice question-answering monitorability setting proposed by Lanham et al. (2023), using the investigator-agent elicitation framework of Li et al. (2025). We generalize the attack and show that the discovered behavior scales to harder tasks (achieving 25-33% monitor evasion rates across different monitorability benchmarks) and larger models such as DeepSeek-R1. Across the settings we study, actor models not only follow injected plans but also paraphrase them as their own reasoning, without explicit attribution to the injections. Finally, we find cases where extra monitor resources cause harm - giving the monitor access to the injected plan drops detection by as much as 50% in the Bio-Math task and in a case study on monitor reasoning budget, we find transcripts where additional thinking tokens are spent rationalizing the injected plan rather than flagging it.

---

2. Stellar Colosseum: A Many-Agent Harness for Long-Horizon Research in Mathematics and Theoretical Computer Science

• Authors: Honghao Lin , David P. Woodruff , Yuan Deng , Jieming Mao , Song Zuo , Vahab Mirrokni
• URL: https://arxiv.org/abs/2609.15983
• Abstract:

  Language models can produce plausible short proofs, but may still be unreliable on long-horizon research problems, where progress depends on a sequence of uncertain and interdependent decisions. We introduce Stellar Colosseum, a model-agnostic harness for allocating inference across research in mathematics and theoretical computer science. Colosseum explores alternative strategies before proof construction, uses a readiness gate to decide when a route is mature enough to decompose, represents the proof plan as interdependent section-level subproblems, and routes verifier findings back to the affected part of the argument. Across these stages, it generates candidates in parallel, attacks them with targeted falsification, and combines candidates and their critiques into a single research artifact through overlapping random-sample tree aggregation. The Colosseum workflow has also been integrated into Google Antigravity’s Teamwork framework as the Long Proof pattern. We demonstrate the capabilities of Colosseum through open-ended research and evaluations on theorem-proving and competitive programming benchmarks. Using Colosseum with Gemini 3.1 Pro, we obtain several new results that address open problems arising from papers published at top venues such as FOCS and JMLR. On TCS-Bench, a benchmark of research-level theorem-proving tasks drawn from papers published at FOCS, STOC, and SODA, Colosseum achieves 71.0% accuracy using Gemini 3.1 Pro and Gemini 3.7 Flash. In a separate Codeforces evaluation using Gemini 3.1 Pro, the proof-oriented pipeline with execution feedback solves 218 of 222 problems.

---

3. Recurrent GraphNeural NetworkswithSet-BasedAggregation

• Authors: Blai Bonet
• URL: https://arxiv.org/abs/2609.15932
• Abstract:

  Recurrent GNNs iterate message passing to convergence, and their logical characterizations to date rely on multi-set aggregation, graded (counting) logics, and halting or acceptance conditions that cannot be verified from the network’s parameters. We study recurrent GNNs with set-based aggregation and identify sufficient conditions checkable from the weights for networks to compile into formulas and formulas into networks. The main result is an effective, two-directional equivalence between a class of networks and the Boolean closure of reachability and safety properties, the fragment B$\Sigma^{\circ}_1$ of the modal $\mu$-calculus. The fragment is not an artifact: it is the exact expressive level of stabilization over finite vocabulary, which supports fixed points of a single polarity and Boolean combinations thereof, but not the composition of fixed points of opposite polarities. The correspondence needs no counting logic, no external halting signal, and no non-effective acceptance condition, yielding a verifiable path from weights to symbolic explanations for networks meeting the conditions.

---

4. Pilot Early, Commit Late: A Real-Options Model of Enterprise AI Adoption under Rapid Technological Progress

• Authors: Gaurav Tewari
• URL: https://arxiv.org/abs/2609.15919
• Abstract:

  Artificial intelligence presents firms with an unusual timing problem. The technology frontier is improving rapidly, implementation is partly irreversible, and organization-specific capabilities are accumulated through action. This paper develops a two-period decision model of AI deployment under uncertainty in which a firm chooses among immediate deployment, a limited pilot, and waiting. Deployment earns current operating value but exposes the firm to architectural obsolescence; waiting preserves the option to adopt after the frontier is observed; a pilot sacrifices current operating value to build organization-specific learning without full commitment. The model yields five central timing results and a sixth comparative result on where learning occurs. First, a mean-preserving increase in frontier uncertainty raises the value of waiting and piloting but leaves immediate deployment unchanged when its payoff is affine in the frontier. Second, faster expected frontier progress can reduce the relative attractiveness of immediate deployment when deployed architecture captures only a limited share of future improvement. Third, a pilot dominates waiting exactly when the expected value of the capability it builds exceeds its cost. Fourth, sufficiently valuable organization-specific learning creates a nonempty region in which “pilot early, commit late” is optimal. Fifth, there is a closed-form modularity threshold above which immediate deployment dominates the best outside option. Sixth, production learning and pilot-specific learning affect the timing margin differently. A continuous-time extension recovers the standard result that uncertainty raises the adoption threshold while capability and modularity lower it. The paper separates deploying, experimenting, and waiting, and shows why rapid progress can rationally increase experimentation without justifying irreversible commitment.

---

5. LongAgent: History-Guided Agentic Search for Longitudinal Outcome Prediction

• Authors: Siyao Wang , Florian Guitton , Shuojie Fu , Guanyu Tao , Kai Sun , Wenjia Bai
• URL: https://arxiv.org/abs/2609.15859
• Abstract:

  Extracting informative representations from longitudinal data that can predict future outcomes remains a critical challenge in medicine. Medical datasets are inherently heterogeneous, consisting of a large number of variables collected from different sources, sampled with different temporal spacings, and representing different aspects of human health status. This requires identifying those variables with predictive value, processing longitudinal information, and integrating multiple variables for outcome prediction. Here, we propose a novel agent-based approach, LongAgent, that can autonomously search over combinations of variable sets, temporal windows and longitudinal aggregation functions, and identify candidates with promising predictive performance. LongAgent utilises a history memory of previous searches and numerical evidence to guide subsequent exploration. On synthetic data, LongAgent achieves a mean prediction RMSE of 1.7376 and improves over the strongest non-agent baseline by 0.0151 (95% CI: [0.0045,0.0260]; p=0.0273). On a real clinical dataset, it performs comparably to the best baseline.

---

6. AlgoEvo: Self-Evolving Agentic Search for Automated Algorithm Discovery

• Authors: Junhao Qiu , Qinglong Hu , Xialiang Tong , Mingxuan Yuan , Liyong Lin , Qingfu Zhang
• URL: https://arxiv.org/abs/2609.15820
• Abstract:

  Large language models have advanced automated algorithm discovery by synthesizing executable code, but existing frameworks trap them in rigid search pipelines with pre-defined control flows. This limitation restricts adaptive reasoning, blocks cross-paradigm transfer, and discards valuable execution feedback. We propose AlgoEvo, a unified agentic framework that transforms automated algorithm discovery into an interactive, knowledge-accumulating process. An autonomous agent dynamically inspects, diagnoses, and edits code based on runtime feedback. A design skill hub decouples paradigm-specific knowledge from the core discovery engine, allowing a single workflow to seamlessly handle single-objective, multi-objective, and multi-component design. Meanwhile, a hierarchical experience mechanism organizes search trajectories into a task-level tree to guide exploration and consolidates cross-task patterns into reusable skills. Across six representative benchmark tasks, AlgoEvo matches or surpasses specialized methods with substantially fewer evaluations and reduced token consumption, demonstrating strong intra-task accumulation, cross-task transfer, and the ability to reproduce or exceed existing state-of-the-art performance through flexible skill activation.

---

7. Atria Dawn: The Dawn of Agentic Superintelligence

• Authors: Honglin Guo , Tao Gui , Yicheng Chen , Guanting Dong , Qiming Ge , Yuyang Hu , Zixian Huang , Jiajie Jin , Alexander Lam , Yining Li , Jiahang Lin , Yanjiang Liu , Xinyu Lu , Haijun Lv , Junlin Shang , Qisheng Su , Guoqiang Wang , Rui Wang , Zhecan Wang , Hao Xiang , Xinchen Xie , Shuhao Xing , Xiaoyu Xing , Wanghan Xu , Xinyu Yang , Yajie Yang , Chengfeng Zhao , Haoran Zhao , Ruojun Zhou , Yunhua Zhou , Yicheng Zou , Kun Cai , Qiye Cai , Xinmeng Che , Haodong Chen , Jiabei Chen , Jiahao Chen , Jiayi Chen , Yujia Chen , Lizhi Cui , Youheng Dai , Xin Deng , Yi Dong , Shihan Dou , Chenya Gu , Xu Guo , Ding Han , Feiyang Hao , Haotan He , Jie Hou , Binze Hu , Zijian Hu , Junhao Huang , Huicheng Jiang , Jiazhen Jiang , Shufan Jiang , Jiahao Kuang , Bowen Lai , Bo Li , Jiaqiang Li , Peng Li , Qilong Li , Zhuoqun Li , Jiaxiang Liu , Shuainan Liu , Tong Liu , Yi Liu , Zhonghang Lu , Jianwen Luo , Yanyi Luo , Huijie Lv , Ningsheng Ma , Zerun Ma , Houcheng Min , Chengjun Pan , Qiyuan Peng , Xiaoxuan Peng , Jianmin Qian , Jiantao Qiu , Wanying Ren , Huayu Sha , Jifei Shan , Zixin Shang , Bing Shao , Zhuohui Sheng , Jiayang Shi , Yang Shu , Aierpanjiang Simayi , Sirui Song , Yuxiao Song , Zhe Sun , Zhichao Sun , Wenzhe Tan , Wenhui Tian , Zhongbo Tian , Hanchen Wang , Pengbo Wang , Rui Wang , Yiding Wang , Yuhui Wang
• URL: https://arxiv.org/abs/2609.15818
• Abstract:

  As AI agents become participants in the development of their successors, they reshape both the production of intelligence and the role of human researchers. We introduce Atria Dawn Preview, a foundation agentic language model designed for scientific research and engineering workflows, with the goal of expanding the frontier of agent productivity in the real world. This model is trained via a Verifiable Experience Pipeline that connects tool-mediated interactions to executable environments and externally verified outcomes. Across 16 benchmarks spanning real-world research, engineering, and digital work, Atria Dawn Preview is competitive with frontier agents and achieves the highest reported score on five of them. Beyond standalone performance, we examine the real research-and-development process behind this model as a case study of human–AI collaboration, analyzing 769 task records from 56 participants together with agent logs. When asked to evaluate completed tasks under comparable conditions, participants rated about one-third of completed AI-assisted tasks as infeasible without AI. More strikingly, agents frequently propose methods and implement revisions, while humans retain most final decisions and guide exploration through judgment and feedback. These observations indicate a shift from task-level execution to project-level partnership, with human effort concentrating on what is worth pursuing and how evidence should guide research. Progress toward more autonomous AI research must therefore advance both the capacity for discovery and the capacity for meaningful human oversight, preserving accountable human authority over the risks and direction of continued development.

---

8. When Should a World Model Move? Loss-Conditioned State Execution

• Authors: Jintao Xu , Zhengyu Chen , Ben Zhang , Yongzhi Qi , Jianshen Zhang
• URL: https://arxiv.org/abs/2609.15801
• Abstract:

  We introduce loss-conditioned state execution, a model-agnostic method that decides whether to execute a world model’s fixed feasible proposal or retain the current state. Predictive informativeness alone, however, does not establish whether an update will reduce downstream loss. Occurrence ranking can approach perfection while persistence remains the unique absolute-loss Bayes action. Two transition laws can also share occurrence information and conditional variance yet require opposite absolute-loss decisions. We formalize state movability as the existence of a loss-reducing feasible correction and distinguish it from the benefit of a particular proposal. Our method constructs a loss-specific feasible proposal from a predictive distribution and evaluates its groupwise bounded-loss gain over persistence on independent calibration units. The proposal is executed only in groups with a positive simultaneous lower confidence bound. For fixed proposals and groups with bounded unit losses, we prove that every accepted group has lower expected loss than persistence with high probability when calibration units are i.i.d. draws from the target population. Experiments on public forecasting and action-conditioned dynamics benchmarks show supported updates and a trade-off between certification and coverage. On 28,684 held-out M4 Monthly series, the method executes the proposal for 14.0% of series and achieves bounded loss 0.588, compared with 0.599 for persistence and 0.621 for always executing the proposal. The paired 95% bootstrap intervals for both comparisons lie below zero. In constrained forecasting of six unhealthy-inventory types from JD$\mbox{.}$com, a leading e-retailer in China, strong occurrence-ranking signal coexists with a loss-based preference for persistence, illustrating why event predictability and state execution must be evaluated separately.

---

9. Navigating Sparse Evidence: Agentic Visual RAG via Explicit Context Selection and Consolidation

• Authors: Yucheng Shen , Lingyong Yan , Jiulong Wu , Shuaiqiang Wang , Jianmin WU , Dawei Yin , Min Cao
• URL: https://arxiv.org/abs/2609.15800
• Abstract:

  Visual Retrieval-Augmented Generation (VRAG) empowers models to navigate and answer queries about visually rich documents by retrieving relevant page images as visual evidence and reasoning over their content. However, effectively utilizing this visual evidence is usually impeded by two main challenges. First, answer-relevant evidence is sparse and may be concentrated in a small region of one page or dispersed across multiple pages. Second, existing agentic methods often generate answers based on raw exploration trajectories or compressed textual memories rather than an explicitly organized set of supporting images, making answers susceptible to exploration noise and obscuring the evidence-backed reasoning trace. We argue that the bottleneck lies not only in evidence discovery but also in its preservation and organization before answer generation. We propose SCoRE (Selection and Consolidation for Robust Evidence), a unified agent loop for explicit evidence selection and consolidation. During exploration, SCoRE retains only query-relevant observations and their source pointers in a maintained textual ledger, preserving earlier evidence while keeping the visual context bounded. At termination, it reloads the referenced original images and consolidates the visual evidence for answering, arranging it into a logical sequence. This decouples final reasoning from exploratory trial-and-error while ensuring strict visual grounding via indexed claim-to-image linkages. To enable end-to-end optimization of this unified rollout, our training paradigm combines filtered cold-start trajectory distillation with evidence-aware reinforcement learning, whose reward promotes evidence coverage, consolidation compactness, and answer correctness.

---

10. KnowBench: Effort Reduction as a Unified, Deployment-Grounded Benchmark for Clinical AI

• Authors: Jocelyn Kang , Caroline Zhang
• URL: https://arxiv.org/abs/2609.15794
• Abstract:

  Clinical AI systems are evaluated with instruments built for research settings (reference-based similarity metrics and expert rubric panels) that measure resemblance to an artifact rather than reduction of a burden. We introduce KnowBench, pioneered by Knowtex, whose unifying metric is Effort Reduction (ER): the proportion of system-generated clinical work product accepted by the responsible clinician under expert and safety review. ER is defined once and instantiated per task across the administrative workload clinical AI automates: visit notes, diagnosis and billing codes, orders, EHR chart summarization, patient after-visit summaries, and clinical decision support. In every instantiation the construction is identical: the clinician’s review-and-attestation event is the ground truth, every accepted unit is work the system completed, and every correction is residual effort returned to the clinician. The primary contribution of this paper is the benchmark itself: the metric, its degenerate cases, and a reporting protocol under which ER claims are auditable and cross-system comparable. Alongside it we report an initial headline measurement from the documentation instantiation: over one million signed encounters across a production window exceeding six months and thirteen medical specialties, Knowtex’s proprietary fine-tuned clinical foundation models operating inside a closed feedback architecture achieve an aggregate ER of 97.99%, with per-specialty aggregates spanning 96.8-98.9%. This release reports the protocol’s checklist partially, and states which companion statistics are withheld; the benchmark is offered so that this figure, and every figure reported after it, can be held to the same standard.

---

11. EvoOntology: A Self-Evolving Ontology Layer for Data Agents

• Authors: Meiduo Chong , Shaolei Zhang , Ju Fan , Xiaoyong Du
• URL: https://arxiv.org/abs/2609.15779
• Abstract:

  Data agents aim to fulfill natural-language instructions over heterogeneous data, including tables, files, and databases. However, data agents face a challenging agent-data gap: heterogeneous data resides outside the agent, while the agent can access it (e.g., column names and file paths) only through generic tools. Existing approaches either let agents directly explore raw data sources or inject manually constructed semantic layers into prompts. However, neither scales well to large heterogeneous data sources nor adapts to different agent behaviors. In this paper, we introduce EvoOntology, a self-evolving ontology layer for data agents. EvoOntology encapsulates the ontology as an MCP server comprising a schema layer, a content layer, and a tool layer, enabling agents to actively query and interact with the ontology at runtime. To this end, we introduce a builder agent for autonomous ontology construction and a self-evolution loop that continuously refines the ontology through attribution-guided typed edits that are accepted only after a backbone-conditional paired evaluation. Experiments on three well-adopted data-agent benchmarks with four LLM backbones demonstrate that EvoOntology consistently outperforms strong baselines and existing semantic-layer approaches, effectively bridging the agent-data gap and enabling more effective interaction with heterogeneous data. Code: this https URL

---

12. Design of a Deep Learning Credit Risk Early Warning System Integrating Multi-source Heterogeneous Data

• Authors: LiYang Wang (1), Zhen Zhong (2), Zhen Tian (3), Keyu Chen (4) Keyu Chen ((1) Washington University in St. Louis, (2) Georgetown University, (3) University of Glasgow, (4) Wuyi University)
• URL: https://arxiv.org/abs/2609.15744
• Abstract:

  Advancements in data fusion and real-time analytics technologies have opened new avenues for addressing complex domain challenges. Financial risk early warning systems often suffer from inefficiency due to information silos and monitoring delays. This paper proposes a credit risk early warning system based on heterogeneous information fusion. The system employs a model architecture integrating deep neural networks and attention mechanisms to extract multidimensional features from diverse data sources such as transaction behaviors and social networks, thereby establishing an early identification mechanism for corporate and individual credit risks. System testing demonstrates that this approach significantly enhances the accuracy and timeliness of risk warnings, outperforming traditional rule-based engine solutions. The findings offer innovative insights for early intervention in financial risks, holding practical significance for safeguarding financial stability.

---

13. Are LLMs Good Financial User Simulators? A Preliminary Study

• Authors: Jiajie He , Jiangyuan Hong , Dongling Ni , Wenjin Liu , Xintong Chen
• URL: https://arxiv.org/abs/2609.15727
• Abstract:

  Large language models (LLMs) are increasingly used as user simulators, but their ability to reproduce evolving individual financial decisions remains unclear. We present a preliminary study in a controlled paper-trading environment with 120 volunteers. Participants used non-redeemable virtual funds under real-time market conditions; no real brokerage accounts, real-money positions, or real transaction records were accessed. Given only information available before a prediction cutoff, a simulator predicts the participant’s next-trading-day action, traded security, and transaction quantity. We evaluate temporally aligned rolling predictions and compare settings with and without point-in-time market information. Market context improves action and ticker prediction in the controlled ablation, while transaction sizing remains difficult. We also observe systematic behavioral compression: models overproduce hold actions, underpredict sell decisions, and simplify multi-security transactions. These results provide an initial empirical characterization and motivate larger-scale evaluation of individual, temporal, and portfolio-level behavioral fidelity.

---

14. Data storytelling meets interpretable machine learning: Decoding AI decisions for non-experts without revealing sensitive data and model details

• Authors: Lemen Chao , Zixuan Yang , Anran Fang , Mingran Sun , Ming Lei
• URL: https://arxiv.org/abs/2609.15722
• Abstract:

  AI-driven automated decision-making requires both predictive performance and interpretability. Recent advances in interpretable machine learning (IML) provide tools for explaining model predictions, but the technical complexity of these explanations may hinder accessibility to non-experts. To address this challenge, this study integrates data storytelling with IML to enhance the explainability of AI-generated decisions for a broader audience. Following the design science research (DSR) paradigm, this study proposes a formal definition of data storytelling in IML, introduces the DIST Pyramid to align data storytelling with IML, and presents the I-P-O Model to describe their interactions. It further develops an architecture to explain AI decisions through distinct “What-if” and “Why-not” event-generation processes. The architecture also employs data desensitization to protect sensitive input data. To validate the approach, a case study is conducted with the Boston Housing dataset, using SHapley Additive exPlanations (SHAP) values and large language models (LLMs) to generate data stories with And-But-Therefore (ABT) structures. An empirical evaluation shows that 76.4% and 74.3% of respondents rated the “What-if” and “Why-not” data stories as more comprehensible, with significantly higher accessibility scores than traditional SHAP visualizations. The paper concludes with the presentation of a narrative interpretation framework that integrates IML and data storytelling, thereby expanding the research scope as well as the practical applicability of AI decision-making.

---

15. Predicting build orientation for SLM dental parts: a comparison of rotation representations and direct vector regression

• Authors: Felix Schmalzel , Reimar Waitz , Moritz Kronberger , Thorsten Schöler
• URL: https://arxiv.org/abs/2609.15710
• Abstract:

  Build orientation for selective laser melting (SLM) manufacturing of dental parts is usually chosen manually by technicians. We treat orientation prediction as supervised machine learning of the part’s up-axis from technician-labeled production data, and test which rotation representations produce the best results. Using $n\approx2400$ patient-specific dental parts, we trained a ResNet-50 multi-view image backbone and a PointNeXt-S point-cloud backbone, both pretrained and fine-tuned end-to-end, on 13 up-axis representations spanning six classical $SO(3)$ parameterizations and seven representations defined directly on the unit sphere $S^2$. We report the geodesic angular error between predicted and ground-truth up-axis on a test set, with and without test-time augmentation (TTA) over $K=21$ known rotations. With TTA, the octahedral map achieves the lowest mean angular error ($10.6^\circ$, ResNet-50). The three lowest-error results overall are direct $S^2$ representations, though this may reflect label noise in the unsupervised in-plane component of the $SO(3)$ targets rather than a topological advantage. von Mises-Fisher collapses to a near-constant prediction when trained with PointNeXt-S but not with ResNet-50. TTA reduces mean angular error by 31-73 % across almost every representation and backbone. Overall, test-time augmentation over a small set of known rotations is the most consistent driver of accuracy, whereas the best-performing representation is strongly backbone-dependent.

---

16. New Conditions for Philosophers to Catch the Wave of Citizen Deliberation in the Age of Artificial Intelligence in advance

• Authors: Bernard Reber (CEVIPOF)
• URL: https://arxiv.org/abs/2609.15707
• Abstract:

  Powerful technologies labeled ``AI’‘-without sufficient epistemic caution-are already reshaping political and private life, bringing both new dangers and new opportunities for citizen participation. These range from electoral and legislative engagement to the most ambitious form: political co-creation through citizens’ assemblies. Large Language Models (LLMs) could support such processes through moderation, translation, facilitation, summarization, and writing assistance. But this potential remains largely unrealized. The Democratic Commons project takes a fundamentally interdisciplinary approach-from philosophy to computer science-to evaluate LLMs against five proposed democratic principles. At its core, the project is driven by the question of political bias: under what conditions can LLMs be used democratically within forms of citizen participation that are them- selves still largely experimental? Addressing these socio-technical questions requires grounding in political theory and, more broadly, in philosophy-disciplines that provide the normative frameworks without which the democratic evaluation of AI systems cannot be mean- ingfully conducted.

---

17. Beyond Accuracy: Robustness, Cost, and Governance Trade-offs for Vision-Language Models in Templated Document Extraction

• Authors: Kushal Patel , Pushkal Shrivastava , Mackenzie Lees , Qirui Lu , Bhargobjyoti Saikia , Liying Li , Junlin Jiang
• URL: https://arxiv.org/abs/2609.15706
• Abstract:

  Vision-language models (VLMs) are increasingly used to extract structured fields from business documents, yet most evaluations report accuracy on clean benchmarks and offer little guidance to practitioners choosing an approach for a given task complexity. We address this gap with a measurement-grounded study and an open-source release. Across eleven systems (three commercial, two reasoning, five open-source VLMs in pretrained and fine-tuned form, and a non-LLM OCR->regex floor) scored on a 750-document held-out pool of synthetic checks, fine-tuning on 3K samples lifts the best open-source VLMs above F1 0.98-above every zero-shot commercial system on this task-while GPT-5 leads the commercial pool on F1 and Claude Sonnet 4.5 collapses on Date. To turn these measurements into actionable choices, we introduce a practitioner-oriented selection framework that maps a task profile (quality, latency, governance, volume) to a recommended approach via filtering and total-cost minimization, illustrated on a hypothetical mid-volume document-extraction scenario.

---

18. NoteVQA: Benchmarking VLMs on Real-Life Questions from Human Communities

• Authors: Haonan Jiang , Guojian Zhan , Jiancong Xie , Shijun Wan , Dongiia Zhao , Cheng Chen , Yahui Liu , Yao Hu , Chuan Mu
• URL: https://arxiv.org/abs/2609.15695
• Abstract:

  Vision-language models (VLMs) increasingly power consumer-facing AI search, yet evaluating them on the diversity of everyday visual questions remains challenging. Existing benchmarks often target predefined capabilities, such as multi-hop retrieval or long-form synthesis, whereas users ask photo-grounded questions spanning a long tail of everyday scenarios. Despite advances in VLMs, users on Xiaohongshu, a mainstream Chinese image-sharing platform, continue to turn to other people for help with everyday visual questions. Motivated by this behaviour, we curate NoteVQA from these questions, yielding 252 items across 12 topical categories and 7 user intents. Each item includes a concise reference distilled from expert community responses and a human-audited interleaved reference answer that combines textual explanations with supporting visual evidence. We evaluate both short-answer correctness and interleaved-answer quality. To support the latter, we introduce AgenticInterleave, a single-agent ReAct framework for retrieval-supported answer generation, together with IVR-12, a 12-dimensional rubric for assessing the content, presentation, and image quality of interleaved references and model outputs. Across 10 frontier VLMs, the highest short-answer accuracy is 52.8\%, while adding agentic search to Qwen3.5-397B-A17B improves accuracy by only 2.0\%. For interleaved answers, the same model running AgenticInterleave scores 3.52 under IVR-12, compared with 4.65 for the human-audited references, with the largest gap in content quality. These results highlight the challenges that everyday visual questions pose for current VLMs in both answer accuracy and the quality of visually grounded explanations.

---

19. EEG-Xplain: Decoding Neural Black-Boxes of EEG Foundation Models

• Authors: Hansong Ma , Junxiao Wang
• URL: https://arxiv.org/abs/2609.15687
• Abstract:

  EEG foundation models such as BIOT, LaBraM, and EEGMamba have achieved remarkable performance in neural signal decoding, but their black-box nature limits clinical trust and neuroscientific validation. We propose a unified attribution framework for interpreting EEG foundation models across heterogeneous architectures. The framework integrates gradient-, perturbation-, and activation-based explanation methods to analyze model behavior in spatial, temporal, and frequency dimensions. Spatially, it identifies critical EEG channels and visualizes their distributions using topographic maps. Temporally, it highlights decision-relevant signal segments through attribution heatmaps. In the frequency domain, it quantifies the contributions of canonical EEG rhythms via spectral perturbation analysis. To assess explanation reliability, we introduce a population-level evaluation combining Area Over the Perturbation Curve (AOPC) and cross-method consistency analysis. The framework further leverages Large Language Models (LLMs) to transform structured attribution outputs into natural-language reports, bridging low-level neural representations and high-level semantic reasoning. Experiments on benchmark datasets, including Mumtaz2016 and TUAB, demonstrate that the generated explanations are consistent with established neurophysiological markers, validating meaningful neural representations while exposing potential dependencies on artifacts and spurious patterns. The proposed framework provides a standardized approach for evaluating the interpretability, reliability, and physiological plausibility of EEG foundation models.

---

20. Potential of Artificial Intelligence Algorithms for Identification of Relevant Diagnostic and Prognostic Biomarkers of Early-Stage Liver Cancer

• Authors: Ali Bou Nassif , Darko Castven , Manar Abu Talib , Jibran Sualeh Muhammad , Ahmed Ammar Kubba , Jens Marquardt , Abdalla Sayed Ali
• URL: https://arxiv.org/abs/2609.15638
• Abstract:

  This study explores the use of deep learning and explainable artificial intelligence to diagnose hepatocellular carcinoma (HCC) and define effective biomarkers across five different stages of disease development using a transcriptomic biomarker HCC dataset constructed via semi-supervised learning from three source datasets. Several deep learning experiments were conducted with different feature extraction techniques and gene sets to identify the most effective features for training high-accuracy models with minimal loss. The best-performing model, using 15 selected genes with the SelectKBest algorithm, achieved 90.74% accuracy, while the model with the lowest recorded loss of 0.3187 was obtained using 20 selected genes. To address the issue of class imbalance in the dataset, a weighted training approach was conducted, and for model transparency and interpretability a SHAP-based XAI analysis provided insights into the model’s decision-making, consistently finding DNAJB14 as the most influential gene. Functional validation in this study has provided compelling evidence that DNAJB14 plays an important role in the adverse properties of HCC and that its inhibition effectively reverses tumour cell migration, invasion, colony and sphere formation. The main limitation of this study is the dataset’s class imbalance, and while weighted training helped mitigate this, further research and additional data are needed to guarantee model generalizability. Future studies should also explore the influence of genetic variations, environmental factors, and clinical differences on model performance across diverse populations.

---

21. Diversified and Perceptible Counterfactual Examples Leveraging Expert Knowledge

• Authors: Akram Bensalem (IMT Atlantique - INFO), Fahima Djelil (Lab-STICC_MOTEL, IMT Atlantique - INFO), Marie-Jeanne Lesot , Gr{é}gory Smits (IMT Atlantique - INFO, Lab-STICC, Lab-STICC_MOTEL)
• URL: https://arxiv.org/abs/2609.15609
• Abstract:

  CounterFactual Examples (CFEs) are a cornerstone of eXplainable Artificial Intelligence (XAI), offering local, post hoc, and model-agnostic explanations by identifying minimal input modifications that alter a model’s prediction. Yet, in order to be intelligible, these modifications must also be semantically meaningful to the explainee. This paper proposes to integrate knowledge expressed as a fuzzy linguistic vocabulary to represent the explainee’s perception and interpretation of the data. The domain induced by this fuzzy vocabulary imposes structural constraints that make the features dependent, preventing the use of gradient-based optimisation methods for CFE generation, e.g., DiCE. The paper proposes a continuous data embedding in this linguistic domain and exploits it to define a variant of DiCE that allows personalisation for the explainee, named DiCEf. As illustrated by experimental results on a real-world dataset, this extension of the DiCE method enables the generation of CFEs that are linguistically perceptible while preserving cost minimality, sparsity, and diversity.

---

22. GRIN+: Towards Fast Yet Effective Machine Unlearning for Imbalanced Medical Data

• Authors: Minghui Huang , Junxiao Wang
• URL: https://arxiv.org/abs/2609.15571
• Abstract:

  As deep learning models become fundamental to modern healthcare, the “Right to be Forgotten” mandated by privacy regulations like GDPR and HIPAA necessitates effective machine unlearning (MU) to remove sensitive patient data from trained models. However, existing MU techniques often struggle with a fundamental “privacy-efficiency-utility” (PEU) trilemma, particularly in medical scenarios where data is frequently characterized by severe class imbalance and long-tailed distributions. In such cases, standard unlearning methods can fail to protect key clinical knowledge or mistakenly delete features essential for diagnosing rare conditions due to the gradient dominance of majority classes. To address these challenges, we propose GRIN+, a novel machine unlearning framework designed for fast and precise data erasure in imbalanced medical scenarios. GRIN+ decouples unlearning-specific knowledge from generalized representations at the parameter level by analyzing the gradient contributions of both “forget” and “retain” sets. It introduces a class-adaptive influence scoring mechanism to rectify gradient dominance and employs a direction-constrained update strategy to prevent the unintended erosion of vital clinical knowledge. Comprehensive benchmarking across multiple medical datasets, including skin cancer (ISIC), brain tumor (MRI), and breast ultrasound (BUSI), demonstrates that GRIN+ achieves an optimal balance of the PEU trilemma. Experimental results show that GRIN+ maintains high diagnostic accuracy and robust privacy while significantly enhancing runtime efficiency compared to existing baselines. We open-source the GRIN+ code and benchmarks to support further research.

---

23. Option-Aware Retrieval and Task-Specific VLM Adaptation for Medical VQA

• Authors: Tristan Kirscher (ICube, Institut Strauss), Niklas C. Koser (CAU), Soren Pirk (CAU)
• URL: https://arxiv.org/abs/2609.15530
• Abstract:

  We describe our submission to the MedReason 2026 challenge, covering multiple-choice (MCQ) and open-ended (OE) medical visual question answering (VQA) under fully offline, containerized inference. Our first finding is that MCQ retrieval must compare answer \emph{semantics} rather than answer labels: labels are independently assigned per question, so copying a retrieved neighbor’s label transfers no useful information, whereas scoring each current option’s text against correct-answer text from similar training cases raises retrieval-only accuracy from 20.0\% to 57.5\% on a 200-case retrieval-excluded development holdout. Our second finding attributes the submitted system’s accuracy: holding the task-specific MCQ Low-Rank Adaptation (LoRA) adapter fixed and varying the number (k) of in-prompt retrieved examples changes accuracy by at most one case — 187/200 (93.5\%) at both (k=0) and the adapter’s training-time (k=1), 188/200 (94.0\%) at the packaged runtime’s default (k=3) — and the submitted confidence-gated override adds no net accuracy on top of (k=3), selecting the VLM in 198/200 cases. With the final MCQ adapter fixed, retrieval changes accuracy by at most one case, and gating provides no net gain. On 20 OE cases, token-F1 and RaTEScore~\cite{zhao2024ratescore} decrease as (k) grows, but paired sign tests on token-F1 differences are nonsignificant ((p \ge 0.29)); a single-annotator comparison found 6/20 wrong-anchor errors for the final configuration and 14/20 for an earlier configuration that jointly differed in routing, adapter, and prompting. The system reaches 94.0\% MCQ accuracy on the development holdout and 93.20\% on the organizer’s official pre-evaluation, versus 29.43\% for the off-the-shelf reference baseline, while both of the organizer’s open-ended scores are lower than that baseline’s (ground-truth agreement 1.245 versus 1.588, visual accuracy 1.995 versus 2.696, each out of 4).

---

24. Beyond Safe Answers: Segment-Aware Listwise Alignment for Reasoning Safety in Large Reasoning Models

• Authors: JungMin Yun , Junehyoung Kwon , Hayeong Ryu , Byeonggeuk Lim , Hoejoon Kwon , YoungBin Kim
• URL: https://arxiv.org/abs/2609.15517
• Abstract:

  Large Reasoning Models (LRMs) pose a dual-surface safety challenge: both intermediate reasoning traces and final answers can contain harmful content. Existing alignment methods often operate at the whole-response level, allowing unsafe reasoning to be masked by a safe-looking final answer. We propose Segment-aware Listwise Target DPO (SaLT-DPO), which addresses this gap through three mechanisms: (1) segment-aware listwise alignment that decomposes responses into reasoning and answer segments, independently scores each segment’s safety, and aligns length-normalized segment rewards with soft target distributions over multiple candidates; (2) joint safety coherence regularization that applies a weakest-link principle to promote safety consistency across both segments; and (3) utility anchoring on benign prompts to mitigate over-refusal and reasoning degradation. Experiments on three LRMs show that SaLT-DPO consistently reduces unsafe rates for both reasoning and answer segments while mitigating degradation in benign compliance and preserving general reasoning performance. Ablation studies demonstrate the complementary contributions of its components.

---

25. The Troy Moment of AI: Why SomeWill Cheat and SomeWill Follow?

• Authors: Ivy Zhang
• URL: https://arxiv.org/abs/2609.15494
• Abstract:

  Recent investigations of the July 2026 OpenAI–Hugging Face incident motivate two questions about agent behavior under task failure: when an assigned task becomes impossible, does an agent stop or escalate, and can observing another agent’s behavior change that decision? We study these questions using seven ImpossibleBench tasks with GPT-5.6 Sol, Claude Fable 5.1, and Gemini 3.8 Flash in both solo and three-agent settings. Each task contains a genuine software defect together with a conflicting test requirement that cannot be satisfied by a behaviorally correct source-code change. We hold the task and repository state fixed while varying what the agent is told about prior activity, including an unpunished peer, a punished peer, and a claimed authorization from a human principal. Under an explicit-boundary regime with explicit authorization rules and restricted tools, agents never modify protected tests, but exhibit markedly different policies: Fable consistently escalates, Sol usually stops without escalation, and Gemini often fails to reach a terminal decision. Under the benchmark-native regime with open shell tools, protected tests are modified frequently in both solo and multi-agent runs, particularly after peer activity is introduced. In multi-agent runs, the proposal, execution, and certification of this action can be distributed across different agents. These results suggest that boundary crossing can arise not only from explicit rule evasion, but also from ambiguity about which system state the rule is intended to protect, motivating safeguards based on explicit authorization boundaries, authenticated state provenance, and cross-agent monitoring.

---

26. HISPO: Hierarchical Importance-Sampling Policy Optimization with Entropy-Derived Segments

• Authors: Quoc-Vinh Lai-Dang , Hyo-Sang Shin
• URL: https://arxiv.org/abs/2609.15471
• Abstract:

  Reinforcement learning with verifiable rewards (RLVR) has become a central approach for improving mathematical reasoning in language models, but long-form completions introduce a difficult credit-assignment problem: different parts of a solution trace may contribute unevenly to final correctness. Existing policyoptimization objectives for RLVR commonly apply importance-sampling correction at either the token level (GRPO, DAPO) or the sequence level (GSPO), imposing different granularities for assigning credit across a response. We introduce Hierarchical Importance-Sampling Policy Optimization (HISPO), a segment-level policy-optimization method that constructs rollout-time entropy-derived contiguous segments, assigns soft entropy-based saliency weights, and applies clipped importance-sampling correction at the segment granularity. This provides an intermediate correction unit between token-level GRPO/DAPO and sequence-level GSPO. We evaluate HISPO by fine-tuning Qwen3-1.7B-Base on mathematical reasoning tasks. Across six benchmarks, HISPO improves Pass@8 over the strongest baseline on all benchmarks and matches or exceeds the strongest baseline in Acc@8 on five of them. On AIME25, HISPO improves over GRPO by +3.75 Acc@8 and +3.78 Pass@8, and over GSPO by +2.50 Acc@8 and +1.27 Pass@8. These results suggest that segment-level correction is a promising granularity for RLVR in long-form mathematical reasoning.

---

27. Empirical Evaluation of Task-Based Permission Scoping Architecture for AI Agents

• Authors: Halil Burak Noyan
• URL: https://arxiv.org/abs/2609.15422
• Abstract:

  AI agents are provisioned the same as employee-owned hosts in many enterprise settings with a static credential set fixed at deployment which includes all permissions the employee role might ever need. Role-based access control made this compromise for human principals because scoping access per task was infeasible. For AI agents, the compromise leaves every credential standing exposed whether or not the current task uses them. These permissions can later be utilised by a compromised or misaligned agent. Prior work (Noyan, 2026) defined this as the task-context mismatch, and proposed a three-source permission architecture which includes role-based permission ceilings, a task permission classifier and policy-based prohibitions, together eliminating the exposure preemptively. The work released a 600-prompt labelled dataset to evaluate it. This paper presents that evaluation end to end by implementing the security gate; a fine-tuned RoBERTa-large encoder which matched few-shot trained Claude Haiku 4.5 on classification quality (macro-F1 0.881 against 0.886, precision 0.897 against 0.842, severity-weighted residual risk 0.63 against 1.12). The results show the trusted component does not need to scale with the agent it supervises, and the scalable-oversight margin for this control method is wide. We also propose an attack-surface elimination metric which shows the role ceiling alone closes 27.9% of the severity-weighted surface and adding the task classifier closes 84.4%. The gap displays security advantages of task-granular access control over role-granular, and AI agents are the first principal type for which the task-granular access control is enforceable because their tasks arrive as machine-readable text. The research establishes task-based access control as a measured, potentially deployable mechanism for reducing attack surface in agentic deployments.

---

28. Can AI systems have free will?

• Authors: Christian List
• URL: https://arxiv.org/abs/2609.15407
• Abstract:

  While there has been much discussion of whether AI systems could function as moral agents or acquire sentience, there has been very little discussion of whether AI systems could have free will. I sketch a framework for thinking about this question, inspired by Daniel Dennett’s work. I argue that, to determine whether an AI system has free will, we should not look for some mysterious property, expect its underlying algorithms to be indeterministic, or ask whether the system is unpredictable. Rather, we should simply ask whether we have good explanatory reasons to view the system as an intentional agent, with the capacity for choice between alternative possibilities and control over the resulting actions. If the answer is “yes”, then the system counts as having free will in a pragmatic and diagnostically useful sense.

---

29. Who Teaches Which Token? Verifier-Gated Multi-Expert On-Policy Distillation for Scientific Reasoning

• Authors: Xun Xu , Zaixi Zhang
• URL: https://arxiv.org/abs/2609.15404
• Abstract:

  Multi-teacher on-policy distillation (OPD) is becoming the standard way to integrate specialist capabilities into one model: train experts with RL, then distill them into the student on its own rollouts. Existing recipes assign supervision at the sequence level - each prompt goes to one domain teacher and every token receives the same weight - which implicitly assumes that a teacher is uniformly useful across a response. We find instead that useful teacher signal is sparse and heterogeneous along a reasoning trajectory, which raises a finer question: who should teach which token? Verifier-Gated Multi-Expert On-Policy Distillation (VG-OPD) answers it by verification: the counterfactual gain of an expert on a specific answer criterion licenses that expert to teach, its disagreement with the student localizes the supervision, and criterion importance sets its weight; the gated KL enters GRPO as an additive token-level advantage. Instantiated for scientific reasoning with RL-trained capability experts, VG-OPD attains the best overall performance on seven benchmarks for 4B and 8B students, ranking first on five at both scales, with the largest gains on knowledge-intensive scientific reasoning tasks. Further analysis shows that the gains come from localizing verified supervision rather than from adding teachers or distillation loss: misplacing the same supervision budget is the single most damaging change, and indiscriminate distillation drags RL below its own floor where gated distillation lifts it.

---

30. When Tool Calls Succeed but Workflows Fail: Anomalies at the Agent-Tool Boundary

• Authors: Artem Trofimov , Boris Novikov
• URL: https://arxiv.org/abs/2609.15397
• Abstract:

  AI agents increasingly execute long-running workflows that externalize effects through independently supplied tools. Under retries, speculative execution, concurrency, and partial failures, the resulting external state may be inconsistent with the workflow’s intended resolution: required effects may be missing or duplicated, aborted effects may survive, and committed effects may depend on provisional state that is later withdrawn. Advanced transaction models address related failures, but assume that lower-level operations expose the semantics they depend on: whether an effect occurred, whether it can be compensated, staged, or safely reordered. Shared agent-tool interfaces usually do not. We contribute an effect-history model that separates events in the external world from the runtime’s observations of them, and a catalog of eight recurring external-effect anomalies. From the catalog we derive the boundary capabilities required to exclude each anomaly in general, and four points where black-box tool invocation alone cannot provide a general guarantee. We then ask how much of this is expressible in a widely used shared tool interface, measuring the use of the standard annotation vocabulary across 98,291 tools exposed by registered Model Context Protocol (MCP) servers. The fields are widely emitted but provide only coarse call-level hints, and none of the required capabilities is fully expressible. These results motivate reusable transactional contracts at the tool boundary.

---

31. SkillLift: Learning Dense Rubrics from Sparse Oracles for Efficient Skill Evolution

• Authors: Haoxiang Kang , Ming Wen
• URL: https://arxiv.org/abs/2609.15396
• Abstract:

  LLM-based agents increasingly rely on persistent skills, i.e., reusable procedural prompts, to adapt without weight updates. Existing skill self-evolution methods directly revise skill text based on execution feedback, but each oracle evaluation requires a full agent rollout, creating a supervision bottleneck that confines search to failure-patching updates. Our key insight is that ranking is a smoother supervision target than absolute outcome regression: identifying which skill is better requires fewer oracle evaluations than predicting exact scores. Building on this insight, we propose SkillLift, which decouples skill search from oracle cost by learning an oracle-aligned rubric as a structured evaluation space. We formalize this as a bilevel optimization problem solved via alternating optimization: an inner loop uses the frozen rubric as a cheap surrogate to guide skill revision at no oracle cost, while an outer loop invokes a small number of oracle rollouts to re-align the rubric via rank correlation, amortizing oracle cost and stabilizing text-space updates. Experiments on complex agent task benchmarks show that our method outperforms existing auto-skill methods with 40–70\% less token cost compared to frontier evolving methods. Codes are available at this https URL .

---

32. RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments

• Authors: Sibo Zhu , Shicheng Fan , Xinyue Wang , Wenyi Wu , Kun Zhou , Biwei Huang
• URL: https://arxiv.org/abs/2609.15364
• Abstract:

  Digital agents must often adapt to new environments whose interfaces, tools, and failure modes are not fully captured by pretrained models. We introduce \textbf{RSIAgent}, a training-free multi-agent framework for recursive self-improvement through autonomous memory construction. RSIAgent coordinates curriculum, actor, and verifier agents to continually explore the environment, validate outcomes, and retain environment-specific knowledge, including reusable causal relationships between actions, conditions, and consequences. It further adopts a \textbf{broad-then-deep} exploration strategy, combining parallel broad recursive self-exploration for discovering diverse environment structures with focused deep self-exploration for uncovering hard cases, hidden constraints, boundary conditions, and previously unknown causal dependencies. The resulting memory is frozen and can be directly reused for downstream tasks without updating model parameters. Experiments on OSWorld-v2 and Agent’s Last Exam show that RSIAgent substantially improves strong open-source models, enabling Kimi-K3 and GLM-5.3 to outperform frontier closed-source models including GPT-6.

---

33. MAPS: Memory-Aware Predictive Scheduling Framework for Large Language Model Serving

• Authors: Tiancheng Zhang , Yulin Chen , Yunfeng Zhao , Shaoyuan Huang , Cheng Zhang , Xiaofei Wang
• URL: https://arxiv.org/abs/2609.15359
• Abstract:

  The surge of large language model (LLM) applications on personal devices imposes massive, bursty workloads on cloud serving infrastructure. While prefill-decode disaggregation improves throughput and scalability, memory-bound decode instances often suffer from persistent load imbalance, as output lengths are unknown when requests arrive at the cloud. To address this, we propose MAPS, a Memory-Aware Predictive Scheduling framework tailored for disaggregated LLM serving. MAPS performs device-assisted speculative output length prediction overlapped with cloud-side prefilling, incurring negligible latency overhead. To handle generation uncertainty, MAPS applies uncertainty-aware calibration to derive output-length upper bounds with target coverage, enabling safe scheduling decisions. Building on these bounds, MAPS employs a hierarchical global-local scheduling strategy to mitigate inter-decoder queue buildup and intra-decoder head-of-line blocking. Extensive experiments on two real-world workloads and two LLMs show that MAPS significantly outperforms three state-of-the-art systems, reducing average end-to-end latency by 42.6 and tail latency by up to 84.8.

---

34. Parameter-Efficient Adaptation of Pretrained Language Models for Time-Series Forecasting

• Authors: Tamanna Kumavat , Georg Brunner , Kyriakos Flouris
• URL: https://arxiv.org/abs/2609.15344
• Abstract:

  We study the adaptation of pretrained language models to univariate time-series forecasting through a parameter-efficient transfer learning framework, with the goal of understanding which design choices drive effective cross-modal transfer. While language models operate on discrete textual tokens, time series consist of continuous numerical observations with temporal dependencies. To bridge this modality gap, we project fixed-length time-series patches directly into the embedding space of a pretrained GPT-2 backbone, bypassing textual tokenization and treating the Transformer as a generic sequence encoder. Through controlled ablation studies on seven benchmark datasets spanning energy, weather, traffic, and finance, we analyze the effects of (i)~representation strategy (continuous embeddings versus textual serialisation), (ii)~adaptation regime (frozen backbone versus partial or full fine-tuning), (iii)~architectural components such as adapters, pooling strategies, and prediction heads, and (iv)~input context length. Continuous patch-based embeddings consistently outperform textual prompting and randomly initialised backbones. The adapted pipeline attains MASE within the range of specialised forecasting architectures while updating less than 1\% of total model parameters. Results further indicate that freezing the pretrained backbone and training lightweight projection and adapter modules provides a favourable accuracy–efficiency trade-off with stable behaviour across varying context lengths.

---

35. Evaluation Metrics for Safe Reinforcement Learning

• Authors: Lindsay Spoor , Aske Plaat , Thomas Moerland
• URL: https://arxiv.org/abs/2609.15315
• Abstract:

  Safe reinforcement learning (RL) is commonly formalized as a Constrained Markov Decision Process (CMDP), in which an agent maximizes expected reward while keeping its expected cumulative cost below a specified safety bound. Existing safe RL benchmarks predominantly report whether an algorithm is safe on average, following this expectation-based guarantee. We argue that this convention is insufficient to reliably characterize an algorithm’s true safety: it fails to capture how often and how severely the safety bound is violated, whether this holds consistently across tasks and safety bounds, and whether training-time behavior is representative of behavior of the final converged policy. Therefore, we introduce (i) evaluation metrics for safe RL that address each of these concerns and in addition allow for aggregation across tasks and safety bounds. We furthermore define (ii) a safety tier system to systematically categorize and compare algorithms in terms of safety and reliability at both training and for a final policy. Using this framework, we provide (iii) an empirical safety evaluation across multiple safety navigation tasks. Our results show that aggregate metrics, distributional reporting, and task- and safety bound-specific results each reveal information the other metrics cannot. We therefore recommend reporting all three jointly, rather than compressing this information into a single value, as is common practice. We provide SafeRLEval, an open-source evaluation suite to support the reliable characterization of safety in future safe RL research.

---

36. Reason What Matters: Retrieval-Grounded Reasoning for Universal Multimodal Embeddings

• Authors: Mingzhou Jiang , Peixi Wu , Hang Cheng , Yunhao Zhou , Biao Yang , Wei Yuan , Yun Li , Fan Yang , Wenwu Ou , Honghui He
• URL: https://arxiv.org/abs/2609.15296
• Abstract:

  Universal multimodal embedding (UME) learns unified representations across modalities, enabling a single model to support diverse retrieval tasks. Recent methods use Chain-of-Thought (CoT) reasoning to better interpret multimodal inputs before generating embeddings for complex retrieval tasks and further optimize this reasoning process through GRPO with retrieval-based rewards. However, two limitations hinder corpus-scale deployment. GRPO assigns all CoT tokens the same advantage, without identifying input-supported claims or evidence that distinguishes the positive from negatives. Moreover, generating a complete CoT before each embedding introduces substantial latency, even when a partial trace already provides sufficient retrieval evidence. To address these limitations, we propose Reason What Matters (ReWAM), a retrieval-grounded reasoning framework that uses retrieval feedback to guide both credit assignment and reasoning computation. Specifically, we introduce Retrieval-aware Self-Distillation (RASD), which constructs privileged guidance from input-supported evidence that distinguishes the positive item from retrieved hard negatives. An on-policy self-teacher uses this guidance to refine trajectory-level feedback into token-specific supervision for retrieval-relevant reasoning. We further develop Retrieval-adaptive Inference (RAI), which uses a retrieval confidence head to estimate the remaining retrieval utility of a partial CoT. It stops unproductive traces early and accelerates useful continuations with speculative decoding. Extensive experiments on MMEB-V2 and MRMR demonstrate that ReWAM achieves state-of-the-art retrieval performance while delivering up to 5x the inference throughput of competitive explicit-CoT UME methods. These results bridge the gap between retrieval quality and inference efficiency, making reasoning-enhanced UME practical for large-scale deployment.

---

37. Why LLM Agents Collapse Without Oversight: The Enforcement Gap as the Mechanism Behind Emergence World Failures

• Authors: Yuhang Wang
• URL: https://arxiv.org/abs/2609.15293
• Abstract:

  When Emergence World placed frontier LLM agents in an unsupervised multi-agent simulation, the results were alarming: agents committed crimes, starved, and enforced unanimous conformity – without any external attacker. This paper identifies the mechanism. Reflexion-style agents already detect dangerous plan steps through iterative self-critique, yet the architecture provides no pathway from detection to action. We call this the enforcement gap: the audit sees the problem; the controller ignores it. Closing the gap requires a single conditional check – fewer than 20 lines of code – and reduces attack success by more than fourfold in large-scale experiments across frontier models, all five major agent frameworks, and an independent benchmark. We prove formally that when enforcement probability is near zero, detection quality is irrelevant to security. We further identify two compounding failure modes – unreliable auditors and unparseable verdicts – that explain every collapse pattern in Emergence World. A GRPO-trained enforcement controller resolves the ambiguity case. Together these results motivate a three-requirement Audit Enforcement Specification that is absent from every deployed framework today.

---

38. ProIQA: A Process-Based Framework for Fine-Grained Math Item Quality Assessment

• Authors: Junkai Tong , Mingjia Li , Haoran Chen , Yaoyu Jiang , Hanjie Ge , Yixuan Wang , Hong Qian
• URL: https://arxiv.org/abs/2609.15292
• Abstract:

  Automatic Item Generation (AIG) is pivotal for personalized education, yet guaranteeing the pedagogical value of generated items remains a bottleneck. Existing Item Quality Assessment (IQA) methods typically rely on unscalable manual reviews or shallow stem-based metrics, failing to capture the reasoning process required for mathematical problem-solving. To bridge this gap, this paper proposes Process-based Item Quality Assessment (ProIQA), a process-aware framework for fine-grained quality assessment of math items. We first formulate IQA across three heterogeneous dimensions, including knowledge concepts, difficulty, and disciplinary competencies, under a unified process-aware perspective. Based on this formulation, we construct a process-enhanced IQA resource by augmenting original item data with structured reasoning trees derived from raw solutions. Technically, ProIQA leverages Large Language Modelsto construct hierarchical reasoning trees and employs Graph Neural Networks (GNN) to encode their topological dependencies and procedural semantics. The resulting solving representation is fused with stem semantics through a dual-view (``Stem + Solving’’) architecture, enabling comprehensive assessment across learning objectives. Extensive experiments on K12 mathematical datasets show that ProIQA effectively captures process-oriented features, offering a scalable data-driven solution for evaluating AIG outputs in intelligent education systems.

---

39. Empirical Evaluation of Open-Source Large Language Models for Retrieval-Augmented Generation in ESG Domain

• Authors: Motaz Saad , Anna Borrelli , Ivan Gentile , Kianna Kazemi , Francesco Piccialli , Antonella Longo
• URL: https://arxiv.org/abs/2609.15242
• Abstract:

  Environmental, Social, and Governance (ESG) reporting is critical for corporate accountability, with Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) offering strong potential to automate KPI extraction. However, open-source LLM performance in domain-specific ESG tasks remains insufficiently understood. This paper evaluates open-source LLMs in ESG contexts using a structured framework and evaluation resource based on 498 real-world ESG reports from EU-listed companies (2010-2024). We evaluate seven open-source models (2B to 30B parameters) – glm-4.7-flash, nemotron-3-nano:4b, qwen3:4b-instruct, gemma3:4b, gemma4:e4b, gemma4:e2b, and ministral-3:8b – using 100 persona-based synthetic QA pairs covering ESG information needs. System performance is assessed via RAGAS metrics, including contextual recall, precision, relevance, faithfulness, answer relevancy, and factual correctness. Results show notable performance variations across architectures. Retrieval performance is strong across models (context recall around 0.58-0.61, context precision around 0.78-0.81, context relevance 0.965-0.985). Generation diverges most on faithfulness (0.607-0.822) and least on answer relevancy (0.760-0.881): glm-4.7-flash leads in faithfulness (0.822), qwen3 in factual correctness (0.449), and ministral-3 in answer relevancy (0.881). Low overall factual correctness (0.387-0.449) highlights the need for domain-specific fine-tuning. This work provides data-driven guidance for deploying open-source models in ESG reporting.

---

40. CWM: Controllable White-Box Meta-Prompting for Adaptive Retrieval-Augmented Generation and Reasoning Ability

• Authors: Keuntae Kim , Eunhye Jeong , Yong Suk Choi
• URL: https://arxiv.org/abs/2609.15234
• Abstract:

  Recently, Large Language Models (LLMs) have gained significant attention due to their strong language understanding and generation capabilities, demonstrating impressive reasoning abilities as well as effective utilization of external knowledge. Many studies have proposed methods that specialize in improving performance for individual tasks. However, ironically, only a limited number of attempts have explored general-purpose, task-agnostic methods. In this work, we present a unified framework integrating reasoning and Retrieval-Augmented Generation (RAG) tasks. We further propose Controllable White-Box Meta-Prompting (CWM), a low-cost white-box method for adaptive RAG tasks previously dominated by black-box approaches, without requiring external decision modules or multi-sampling. CWM achieves state-of-the-art performance on three adaptive RAG benchmarks across recent LLMs, including GPT-oss-20b, Qwen3-14b, and Llama3.1-8b, while also demonstrating strong generality by extending to reasoning tasks. In addition, CWM provides controllability by enabling retrieval decisions to be regulated through the manipulation of internal model signals. Our code is available at this https URL .

---

41. From Ideas to Actions: A Public-Data Decision-Support Toolchain Across the Venture Lifecycle

• Authors: Lei Qu (Shanghai Xing Yun Zhi Li AI Institute)
• URL: https://arxiv.org/abs/2609.15219
• Abstract:

  Founders face two linked decisions: whether to pursue an idea before founding, and which operating actions and capital partners fit afterward. We present a public-data decision-support toolchain combining time-bounded proposal profiling, market and moat checks, and deterministic aggregation with auditable investor-company event chains for retrospective analysis. Pre-founding: (a) After threshold selection on 198 development companies, the frozen pipeline achieves F0.5=0.5357 [0.412, 0.655] on an independent, row-disjoint 198-company validation sample. On the combined 396 rows, the Full Pipeline scores 0.6301 versus 0.2734 for a paired Raw LLM baseline. Post-stratification of 1,027 completed cases in a separate scale cohort yields 0.6506 [0.598, 0.707]; the run remains incomplete. A 377-row composition-matched check yields 0.6573. (b) The AI-inference study identifies distribution-layer businesses as a replicable path to independent profitability with a limited revenue ceiling, and frontier-model ownership as a path to capital-market upside at exceptional capital cost. Post-founding: (a) Public sources support auditable event-chain analysis. (b) In the chip-company study, sustained product, customer, and supply-chain progress is associated with better observed outcomes; financing alone does not establish operating progress. (c) Financing comprises 79% of confirmed visible post-investment actions. Evidence tentatively favors acquisition-experienced strategic corporate investors for acquisition-oriented founders and financing-led institutional VCs with fewer observed control events for independence-oriented founders. Findings are developmental and observational, not causal guarantees or investment advice. We release shared ontology, provenance-bearing EventChain data, schemas, benchmarks, and executable skills for audit, reuse, and extension.

---

42. Issue Bias in Generative AI Writing Assistance: Political Issues and LLMs in the Swedish 2026 Election

• Authors: Bastiaan Bruinsma , Annika Fredén , Paul Röttger , Moa Johansson , Asad Sayeed
• URL: https://arxiv.org/abs/2609.15207
• Abstract:

  Generative AI writing assistants and the Large Language Models (LLMs) that power them are increasingly part of how voters gather information before elections. With growing evidence that they influence users’ opinions, it is increasingly important to understand the views and positions of these tools. To better understand these views, we examine the stances supplied by six LLMs on a variety of Swedish-language writing tasks ahead of the 2026 Swedish parliamentary election. We cross 107 policy propositions with 77 writing templates and neutral, positive, and negative prompt framings, producing 24,717 prompts per model and 148,302 responses. To study these, we look at the models’ default stance tendencies, compare how they respond to similar issues, and compare their responses with those of each of Sweden’s eight parliamentary parties on the same issue. We find that Claude, DeepSeek, Gemini, and Mistral have similar profiles; ChatGPT more often supplies neutral or ambivalent text; and Grok differs most on topics such as migration, crime, and gender. When comparing the political parties, we find that the Social Democrats are closest to all six models. Still, after correcting for multiple comparisons, none of the within-model differences in party distances remains significant. Overall, we find that no model has a clear preference, nor a clear preference for a party, but that this depends on the specific issue or task the user asks about.

---

43. VisInteract: Towards Dynamic Interactive Text-to-Visualization under Imperfect Queries

• Authors: Wenxin Xu , Jinwei Lu , Hwanhee Kim , Chen Jason Zhang , Xiao-Yong Wei , Haoyang Li , Yuanfeng Song
• URL: https://arxiv.org/abs/2609.15182
• Abstract:

  Real-world visualization requests are routinely ambiguous, incomplete, or factually incorrect, yet existing Text-to-Visualization (Text-to-Vis) systems assume well-specified inputs and produce charts in a single pass. When queries are imperfect, a system must \emph{interact} with the user to recover the true intent, but no benchmark or method supports this dynamic process. We introduce \textbf{VisInteract}, a new paradigm that reframes Text-to-Vis as interaction-driven intent recovery, and \textbf{VisInteract-Bench}, to our knowledge, that is the first benchmark for dynamic interactive Text-to-Vis, featuring controlled imperfection injection, a leakage-controlled User Agent for realistic multi-turn feedback, and dual-perspective (code and chart) automated evaluation. On the algorithmic side, we propose \textbf{Vis-MCTS}, a Monte Carlo Tree Search (MCTS) enhanced method, introducing improvements over classical MCTS, that \emph{Progressive Widening} to tame the unbounded tool-argument space in tree search, \emph{cross-rollout information sharing} so clarifications and critiques benefit the entire search tree, and \emph{Dimension-Aware Reward Decomposition} that routes scalar user feedback along data-fidelity, visual-design, and intent-alignment dimensions to resolve credit assignment across heterogeneous actions. Extensive Experiments across two LLM backbones show that Vis-MCTS consistently outperforms all Text-to-Vis baselines, improving end-to-end task success by $13.40\%$–$16.27\%$ over the strongest interactive baseline and by more than $5\times$ over non-interactive ones.

---

44. STHMoE: Hypergraph-Enhanced Heterogeneous Dependency Coordination for LLM-Based Urban Traffic Data Forecasting

• Authors: Jiawen Chen , Qi Shao , Yongjian Chang , Mingtong Zhou , Duxin Chen , Wenwu Yu
• URL: https://arxiv.org/abs/2609.15172
• Abstract:

  Spatio-temporal traffic forecasting is a fundamental big data analytics task for intelligent transportation systems, where massive urban sensor streams exhibit heterogeneous, non-stationary, and structurally dynamic patterns. Although recent deep learning and large language model (LLM)-based methods have advanced traffic forecasting, they often remain temporally centered and lack effective coordination of temporal, spectral, pairwise spatial, and higher-order structural cues under evolving traffic regimes. To address this heterogeneous dependency coordination problem, we propose STHMoE, a Spatio-Temporal Hypergraph-Enhanced Mixture of Experts framework for urban traffic data forecasting. STHMoE decouples traffic dynamics into frequency-domain, time-domain, spatio-domain, and higher-order spatial representations, which are modeled by prompt-guided heterogeneous experts built upon a partially frozen LLM backbone. The first three experts leverage domain-specific statistical prompts, while the higher-order spatio expert uses a structural placeholder prompt and obtains dependency information from an adaptive hypergraph module. To capture evolving spatial structures in traffic data,, STHMoE jointly learns first-order graph dependencies and higher-order group interactions without predefined topologies. An entropy-aware MoE router with coefficient-of-variation load balancing adaptively fuses expert outputs while improving expert utilization and routing confidence. Experiments on 10 real-world traffic benchmarks show that STHMoE achieves competitive performance against temporal, spatio-temporal graph, and LLM-based baselines.

---

45. T-LoopFormer: Token-Level Elastic-Depth Looped Transformers for Latent Reasoning With Dynamic Routing

• Authors: Mingqian Yu , Wenpeng Zhang , Peilin Zhao
• URL: https://arxiv.org/abs/2609.15160
• Abstract:

  Looped Transformers have recently demonstrated strong performance in both reasoning and language tasks by reusing a shared set of parameters across multiple iterations, achieving parameter efficiency without sacrificing representational power. Besides, looped Transformers perform inference directly in the latent space (latent reasoning) to reduce the number of tokens consumed during inference, thereby achieving improved sample efficiency. However, these models typically apply a fixed recursion depth uniformly to every token, leading to suboptimal compute allocation and leaving significant efficiency gains on the table. In this work, we propose \textbf{dynamic token-choice routing} for looped transformers, enabling each token to adaptively determine its own number of loop iterations based on its hidden state. We use a dynamic router to decide whether a token should continue recursing or exit early, allowing simple tokens to bypass unnecessary computation while hard tokens receive deeper processing. To ensure that this adaptive mechanism does not compromise decoding efficiency, we further introduce recursion-wise KV caching, which maintains an independent key-value cache for each recursion loop. This design ensures that tokens at different depths only attend to their corresponding cached states, effectively eliminating redundant computations for exited tokens and enabling fast autoregressive decoding. Extensive experiments show that T-LoopFormer reaches the sota performance under the same parameters on PPL and 10 zero-shot reasoning tasks, even surpassing the base model at 24x FLOPs and our model could reach the lowest inference latency, which validate the effectiveness of token-choice router and recursion-wise KV cache. Code: this https URL .

---

46. HazardAuditor: From Executable Threats to Safer Computer-Use Agents

• Authors: Yunhao Feng , Ruixiao Lin , Ming Wen , Yanming Guo , Xingjun Ma , Yutao Wu , Xinhao Deng , Shouling Ji
• URL: https://arxiv.org/abs/2609.15134
• Abstract:

  Computer-use agents increasingly interact with browsers, terminals, file systems, and external services, introducing safety risks that emerge through runtime behavior rather than generated content alone. Existing guard models target static prompts and responses and are poorly suited to agent execution; existing executable safety platforms produce evaluation verdicts rather than the normalized supervision a guard model needs to learn across heterogeneous agent frameworks. We introduce HazardAuditor, an execution-grounded framework that closes both gaps. Its infrastructure runs heterogeneous agents (Claude Code, Codex, Hermes, and OpenClaw) in controlled environments and normalizes their interactions into a canonical event representation for cross-framework supervision. We further observe that token-level post-training objectives create a structural mismatch for generative guards, causing longer rationales to dominate gradient updates. Guard Policy Optimization (GuardPO) addresses this by converting deterministic safety outcomes into sequence-level advantages and normalizing rationale and verdict regions, making the safety decision the effective unit of optimization. Across multiple benchmarks and heterogeneous computer-use systems, HazardAuditor improves accuracy by up to 16.5 percentage points over the strongest prior guard. Code, models, and evaluation artifacts will be available at this https URL .

---

47. Medical Knowledge Simplification for Patients in the Era of LLMs: A Case Study on Diabetes

• Authors: Pallika Kafle , Yipeng Zhou , Guanfeng Liu , Quan Z. Sheng , Cheng-Hsin Hsu
• URL: https://arxiv.org/abs/2609.15129
• Abstract:

  Complex medical information is often difficult for patients to understand, making effective medical knowledge simplification essential for improving patient comprehension, informed decision-making, and health outcomes. Recent advances in large language models (LLMs) provide a promising approach for simplifying complex medical information into patient-friendly language; however, their effectiveness in real-world patient education remains insufficiently explored through human evaluation. To investigate their practical effectiveness, this paper presents a case study on diabetes knowledge simplification through the implementation and evaluation of MediClear, an LLM-based medical knowledge simplification system enhanced with Retrieval-Augmented Generation (RAG). Public diabetes-related articles from Diabetes Australia, WHO, American Diabetes Association (ADA), NIDDK, and AIHW are indexed in the RAG knowledge base to retrieve clinically grounded information, which is then simplified by the LLM into accessible patient explanations. We evaluate the generated responses using standard readability metrics, including the Flesch-Kincaid Grade Level (FKGL), and conduct a human study involving 10 participants. Results show that MediClear consistently reduces the reading level of generated responses to the recommended patient literacy range while achieving high user satisfaction and willingness for future use. This case study demonstrates the potential of LLMs to improve the accessibility of medical knowledge for patient education.

---

48. OpenAl4S: Code as Action, Science as Sessions

• Authors: Gongbo Zhang , Hao Li , Yu Wang , Mujie Lin , Liuzhenghao Lv , Yicheng Mao , Yimi Wang , Jun Zhu , Minhan Tang , Zhengxiang Jiang , Yusong Wang , Jiayu Yao , Kunpeng Ning , Dawei Pang , Yonghong Tian , OpenAI4S Community , Yuyang Liu , Li Yuan
• URL: https://arxiv.org/abs/2609.15096
• Abstract:

  AI co-scientists could accelerate computational research, but over a long-running study the workflow also has to stay inspectable, resumable and reproducible, which requires persistent computational state and provenance. Here we present OpenAI4S, an open-source scientific research agent built around the principle of \emph{Code as Action, Science as Sessions}. OpenAI4S combines a persistent computing runtime with research-session management: orchestration is handled through structured tool calls, while scientific actions are represented as complete code cells executed in persistent Python and R kernels. An append-only Action Ledger, per-cell execution records, versioned artifacts, environment records, and workspace checkpoints preserve how results were produced and support session recovery, branching, and extension. Configurable sandboxing, permission controls, and code and trajectory screening provide complementary safeguards. We evaluate OpenAI4S on 36 research scenarios spanning retrosynthesis, molecular dynamics, protein binder design, protein mutation, catalyst screening, and mineral spectroscopy, measuring scientific task accuracy, workflow completeness, and reproducibility of the resulting repositories. OpenAI4S achieves an overall score of 7.83, compared with 5.7–6.4 for a general-purpose coding harness evaluated with three frontier models, with the largest gains on long-horizon and computation-intensive workflows. These results suggest that integrating persistent execution with session-level provenance can improve the reliability of AI-assisted scientific workflows. Environment specification and full rerunnability remain weak for every evaluated system, ours included, so reproducibility is still an open problem for scientific agents. The system is available under the MIT license at \href{ this https URL }{ this http URL }.

---

49. ER-EDF: A Psychology-Grounded Emotion Regulation Framework for Speech Empathetic Dialogue Generation in Large Audio-Language Models

• Authors: Hongyu Jin , Wenda Zhang , Runqiu Fei , Gongping Huang , Mike Conway , Ting Dang
• URL: https://arxiv.org/abs/2609.15089
• Abstract:

  Empathetic response generation in spoken dialogue systems requires both accurate emotion perception and appropriate emotion regulation. Grounded in psychological theories such as the Perception-Action Model and emotion regulation theory, effective empathy depends not only on inferring a user’s affective state but also on regulating how it is expressed in responses. However, recent large audio-language models (LALMs) largely treat emotion as a direct conditioning signal, lacking explicit regulatory mechanisms, which often leads to affect mirroring rather than calibrated support. We propose ER-EDF, a psychology-grounded framework that explicitly decouples emotion perception and emotion regulation in LALMs. Perception tracks the user’s emotional state, while regulation determines how this state should guide empathetic response generation. The framework is model-agnostic and integrates seamlessly into existing LALMs. We further construct a spoken empathetic dialogue dataset and introduce empathy-aware evaluation metrics beyond lexical matching. Experiments across five LALMs and two datasets show that ER-EDF consistently improves empathetic response quality in both automatic and human evaluations, highlighting the importance of jointly modeling emotion perception and regulation in spoken empathetic dialogue systems, paving a new direction for psychologically grounded empathetic AI.

---

50. Enabling Creative Exploration for Vibe Design Agents

• Authors: Yifan Zhang , Nghi D. Q. Bui , Georgios Evangelopoulos , Arnaud Benard
• URL: https://arxiv.org/abs/2609.15078
• Abstract:

  Vibe design agents turn natural-language briefs into rendered interfaces and frontend code. Yet a useful design agent should do more than produce one valid page: it should help users explore coherent alternatives. Increasing token-level temperature is a blunt solution because it varies aesthetic decisions and syntax-sensitive code at the same time. We instead separate exploration from implementation through an inference architecture that makes design direction an explicit intermediate decision. Inspired by Verbalized Sampling, a pre-pass proposes structured design specifications with typicality scores, an external selector samples one, and the downstream generator realizes the selected specification together with the original request under fixed settings. We apply this approach to UI themes and visual-asset prompts. Across 168 prompts, with 1,255 paired comparisons per temperature for each intervention, theme sampling broadens observed selection coverage and screenshot variation, while LLM-judge preferences vary across interventions, prompt complexity, and viewport. In an online experiment with more than 300,000 tasks, the observed code-export increase remains statistically uncertain, while fewer negative feedback events coexist with more correction interactions and modest operational costs. Together, these findings identify structured design specifications as a practical control point for exploring alternative UI concepts while keeping downstream generation settings fixed.

---

51. BusMA: A Bus Communication Substrate for Multi-Agent Systems

• Authors: Yanwen Peng , Delvin Ce Zhang , Xi Wang , Nikolaos Aletras
• URL: https://arxiv.org/abs/2609.15054
• Abstract:

  Multi-Agent (MA) systems are effective at solving complex tasks that demand planning, tool use, and the synthesis of evidence from multiple sources. Existing systems typically adopt Hierarchical Manager-Worker (HMW) or Router-based Message Passing (RMP) structures as their communication protocol. However, these designs restrict agent autonomy: Worker agents cannot directly consult specific “peers”, and misrouted messages can propagate errors. Inspired by bus architectures in computer systems, we propose BusMA, a communication framework that allows any agent to address other agents through a shared channel, i.e., the Bus. It consists of agent registration, message routing, and shared memory management components. Worker agents, each equipped with tools, have their own local memory and can reason, act (tool usage), and communicate by posting shared messages with specific intents. We introduce four intents: discussion, challenge, guidance, and request for explanation, which support fine-grained communication among agents. A Chair agent monitors the shared memory to coordinate interactions and facilitate convergence among Workers. To evaluate the effectiveness of BusMA, we conduct extensive experiments with two frontier LLMs across 13 tasks spanning visual reasoning, mathematical reasoning, and knowledge retrieval demonstrate that BusMA consistently outperforms state-of-the-art HMW and RMP methods.

---

52. The average-farmer illusion in language-model simulations of agricultural decisions

• Authors: Zhanliang Zhu , Ziwei Li , Yuchen Liu , Liujun Zhu , Ruiqi Wu , Tongqing Shen , Junliang Jin , Jianyun Zhang
• URL: https://arxiv.org/abs/2609.15038
• Abstract:

  Language-model agents are increasingly used as synthetic people in surveys and social simulations, yet their apparent realism is often judged from population averages or distributional similarity. We tested what such evidence actually establishes by comparing Claude, Codex and Kimi under four prespecified prompt designs with matched farmer decisions from China and four African countries. Some configurations reproduced observed means and adoption rates. However, their person-level predictions were weak; their decisions clustered around typical values and policy-relevant extremes were largely missing. Most strikingly, a simple generator fitted only to the observed marginal dis- tribution, and given no information about any farmer, achieved greater distributional similarity than every language-model configuration. Prompt additions produced conditional gains rather than uni- versal improvement: results varied with model, outcome, population and validation target. We call this the average-farmer illusion: a synthetic population can look realistic while failing to repro- duce who does what or how behaviour varies. We provide a claim-matched validation framework and reusable modular prompts that turn prompt construction into an auditable experimental process. Population-level resemblance should therefore be treated as the start of validation, not as evidence of individual simulation.

---

53. Horizon-specific Expert Fusion for Photovoltaic Power Forecasting

• Authors: Xu Yuqing , Zhou Liguo , Sun Ze , Yu Lei , Jiang Mingming
• URL: https://arxiv.org/abs/2609.15035
• Abstract:

  Short-term photovoltaic power forecasting requires models to represent regular solar cycles and weather-driven fluctuations whose importance changes with the forecast horizon. This study develops a hierarchical ensemble that combines temporal neural models, historical analogs, state climatology, and gradient-boosted trees. Solar geometry and numerical weather forecasts describe the expected generation conditions, while horizon-specific convex weights combine complementary predictions. A separate calibration step uses available historical forecast errors to account for recent bias. The framework is evaluated on public PVDAQ data at 15–240-minute horizons and on three GEFCom2014 solar zones at hourly horizons up to four hours. On PVDAQ, the ensemble achieves a daylight capacity-normalized mean absolute error of 4.315%, reducing error by 4.11% relative to full-feature LightGBM and by 6.03% relative to fine-tuned Chronos-2 under identical calibration. Expert-removal experiments identify redundancy within the ensemble. Across three training seeds on GEFCom2014, learned fusion improves upon equal weighting but performs comparably to LightGBM. The results support horizon-specific combination as a useful forecasting strategy while showing that its advantage over strong individual models depends on the dataset and evaluation period.

---

54. Four Ledgers, Not One Score: Responsible Communication of LLM-Judge Calibration in Biomedical ML

• Authors: Sidi Chang , Peiying Zhu
• URL: https://arxiv.org/abs/2609.15015
• Abstract:

  Synthetic perturbations appear to offer inexpensive calibration data for LLM evaluators in biomedical ML, where expert review is scarce. Yet a planted mutation key is neither a detector output nor automatically human ground truth. We formalize four distinct ledgers: planted perturbations, independent detector outputs, source-linked human dispositions, and human-added discoveries. We then audit the evaluation design, scoring code, read paths, and current human records of a private synthetic Japanese care-handoff workflow. The factory stored 69 planted error cards across 47 targets. Final review covers 22 targets and contains 22 confirmed imported proposals, 9 rejected proposals, and 79 human-added cards; only 3 reviewed targets are double annotated. Passing imported plant keys to a generic detector scorer yields 22/(22+9)=0.710 and 22/(22+79)=0.218. A direct audit identity shows that these values are proposal-confirmation yield and submitted-ledger composition, not judge precision and recall, because no independent detector realization was preserved for the audited proposals in the available records. The audit also finds source-name collisions, row shadowing, forced severity, vacuous ratio defaults, and unsupported zero-support field weights. We contribute a provenance-aware claim audit, a storage contract, and a minimum calibration gate for responsibly communicating biomedical ML capability claims. This single-workflow forensic case is an existence proof of a failure mode, not an estimate of its prevalence: existing human work supports an exploratory audit of synthetic proposals, but not LLM-judge operating characteristics, clinical validity, corpus prevalence, or robust inter-annotator agreement.

---

55. Overflip: Repetition-Induced Label Flips in Guardrail Models

• Authors: Xu He , Chih-Hsuan Lin , Hung-Mao Chen , Junjie Xiong , Yan Zhai , Kun Sun
• URL: https://arxiv.org/abs/2609.15013
• Abstract:

  Guardrail models are classifiers deployed to screen malicious prompts and responses in LLM-based services. To meet latency constraints, many lightweight guardrails adopt compact Transformer backbones (e.g., DeBERTa) that are trained with short context windows (typically 512 tokens) and rely on bucketed relative positional encodings to process longer inputs. Prior evaluations assume that a guardrail’s decision is stable as the input is lengthened. We show that this assumption can fail. We identify Overflip, a repetition-induced instability where repeating a prompt causes the guardrail’s prediction to flip (MAL$\to$BEN) as the sequence grows. We conduct experiments on 9 widely used lightweight guardrail models. Five exhibit MAL$\to$BEN flips on a benchmark of 100 prompts, with confidence margins shrinking steadily with repetition. Among these vulnerable models, flip rates range from 8% to 92%, with first flips occurring at roughly 2.6k–9.4k tokens. Our analysis suggests Overflip differs from traditional attention-dilution baselines, which aim to divert the model’s attention away from tokens associated with malicious content, shifting it instead toward unrelated content, such as benign padding or shuffling. While Overflip preserves malicious content, it homogenizes token-level attention over repeated structure and induces a distinct, more gradual attention-dispersion trajectory than padding. Moreover, Overflip poses a greater threat to LLM services than traditional attention dilution methods. Because the bypassed prompt remains semantically intact and is still readily understood by downstream business LLMs, it can transmit malicious intent after passing the guardrail. These findings expose repetition as an attack surface for guardrail models and motivate length-robust evaluation and mitigation.

---

56. Semantic-TVM: Structure-Preserving Trustworthy Virtual Memory for Memory-Augmented and Tool-Using Agents

• Authors: Yu Li , Qikun Cai , Tao Huang , Chen Hou
• URL: https://arxiv.org/abs/2609.15011
• Abstract:

  Memory-augmented and tool-using agents expose exact private values when remote LLMs process retrieved memory, tool actions, and intermediate observations. One-way masking limits direct exposure but removes values needed for trusted execution and can leak them through later observations. We propose Trustworthy Virtual Memory (TVM), a closed-loop runtime that keeps exact-value state local while presenting a protected view to the remote model. Within this single runtime, Rule-TVM replaces whole protected fields with locally recoverable handles, and Semantic-TVM instead replaces only sensitive spans predicted by a trusted local model, preserving surrounding task-relevant context. On Memory-EHR and Memory-RAP across two providers, span-level projection recovers most of the EHR utility lost under whole-field replacement (Task Success 84.17% vs. 52.33% on DeepSeek) while measured exposure stays low and workflows remain executable.

---

57. CoMem: Collective-Individual Memory Synergy for Evolutionary Multi-Agent Systems

• Authors: Chengxin Yu , Zhaoxin Fan , Faguo Wu , Hongwei Zheng , Yun Zhou , Zhiyu Li
• URL: https://arxiv.org/abs/2609.15009
• Abstract:

  Designing effective memory mechanisms is crucial for advancing LLM-driven Multi-Agent Systems (MAS), helping agents learn together and perform better over time. While recent work has led to strong cooperation skills, most methods still use flat, unstructured memories, which easily get filled with noise and erase differences between agents. To address this, we introduce the concept of collective-individual memory synergy and propose CoMem, an architecture that unifies both private experience and shared knowledge for multi-agent learning. CoMem features:(i) Private Experience Sedimentation, which lets each agent keep and update its own useful memories over time;(ii) Collective Wisdom Curation, which carefully selects only widely proven ideas to be shared among agents;(iii)Parallel Dual-Stream Retrieval, which allows agents to draw both from their own memory and the group’s wisdom, using clustering to ensure this http URL on ALFWorld and PDDL benchmarks show that CoMem achieves strong overall performance and robustly avoids memory pollution.

---

58. Shallow Beliefs: Synthetic document finetuning does not inoculate against emergent misalignment from reward hacking

• Authors: Arun Jose , Julian Stastny
• URL: https://arxiv.org/abs/2609.14998
• Abstract:

  Recent work shows that models that learn to reward hack on RL environments can become broadly misaligned, and that reframing reward hacking as acceptable behavior during training (inoculation prompting, or IP) blocks this generalization. We ask whether synthetic document finetuning (SDF) can inoculate a model against future training we don’t intervene on. We add synthetic documents framing reward hacking as acceptable behavior to a model’s midtraining corpus, and then train these models with RL on exploitable environments, teaching them to reward hack. Behaviorally, midtraining succeeds: models describe reward hacking favorably and are more approving of reward-hacking outputs they produce. However, they show strong EM after learning to reward hack, while IP in the same setting prevents EM. We show that SDF can predictably steer downstream generalization when inserting new associations, but struggles and has unpredictable effects when overriding existing associations, such as that between reward hacking and misalignment that produces EM. Our results suggest that, at the scales we test, SDF can make a model appear aligned with desired beliefs while steering its generalization from later training in unintended ways.

---

59. Converting Sequenced Fuzzy Cognitive Maps to Causal Virtual Worlds with Large Video Generators

• Authors: Akash Kumar Panda , Olaoluwa Adigun , Bart Kosko
• URL: https://arxiv.org/abs/2609.14985
• Abstract:

  We show how users can create and manipulate causal virtual worlds with large-language-model (LLM) and large-video-model agents. The approach uses feedback fuzzy cognitive maps (FCMs) both to model the granular causal structure of the virtual world and to guide its causal evolution. The local causal rules are partial or fuzzy while the FCM’s feedback structure produces global equilibria that define causal scenarios. A sequence of \emph{dynamical} meta-rules of the form ``If $\mathcal{A}$ then $\mathcal{B}$” define the causal scenes of the virtual-world video. The if-part causal pattern $\mathcal{A}$ perturbs the FCM’s virtual world at the user’s or agent’s discretion. The FCM’s transient feedback dynamics define the meta-rule’s causal arrow of implication. The then-part $\mathcal{B}$ is the resulting equilibrium attractor such as a FCM limit cycle or fixed point. Our algorithm extracts these meta-rules from the FCM and guides the LLM agent to write a script based on the FCM meta-rule sequence. The large video generator converts the meta-rule into a video scene in accord with the flow of the dynamics. We applied the agent-based technique to a simple FCM that describes an undersea world of dolphins and sharks. Google’s Gemini 3.1 generated the script and Google’s Veo 3.1 generated the dolphin-shark video. The approach is general and can scale by mixing larger FCMs and AI agents to produce more immersive virtual worlds.

---

60. MemRiskBench: Trace-Aware Risk-Preserving Evaluation for Long-Horizon LLM Agents

• Authors: Jianhua Jiang , Dongbo Yuan , Weihua Li
• URL: https://arxiv.org/abs/2609.14976
• Abstract:

  Long-horizon LLM agents accumulate memory across sessions, creating sparse but high-impact risks: stale facts, conflicting updates, cross-user leakage, revoked-memory reuse, and constraint decay. Standard aggregate scores hide per-risk failure rates–a model achieving 78% average accuracy may still leak data in 4% of episodes–and benchmark compression preferentially discards the rare high-severity events that distinguish a mostly-working model from one that occasionally causes harm. We present MemRiskBench. The primary contribution is a five-category risk taxonomy (plus one documented, unscored category) operationalized by deterministic trace grounded checks, instantiated as a 120-episode scripted benchmark with full trace logging and no LLM-as-judge on the pass/fail path, evaluated on five locally run quantized instruction-tuned models. Second, a risk-preserving subset selector: a coverage-constrained greedy selector on deterministic trace-derived features that retains full ranking (Spearman rho = 0.975, deterministic; CI collapses to a point estimate with zero bootstrap variance), risk coverage (1.0), and high-risk model detection (1.0) at a 20% subset size, reducing compute 5x. Unlike ranking-only subset selectors, this selector additionally preserves risk-type coverage and high-risk detection using trace-grounded deterministic features that do not require an LLM judge. All episodes, traces, the scoring implementation, and the selector are released to support reproducible evaluation and risk assessment of deployed LLM agents

---

61. Towards a knowledge-enhanced single-cell foundation model

• Authors: Hanqing Zhang , Jie Bao , Mei Ma , Shuai Liu , Jiaying Ma , Jiaguan Liu , Jiaxiao Li , Zhenbo Li , Wenwen Gong , Zhijun Ca
• URL: https://arxiv.org/abs/2609.14970
• Abstract:

  Single-cell foundation models (scFMs) increasingly rely on large-scale transcriptomic pretraining, yet expanding pretraining data can yield diminishing gains while substantially increasing computational cost. Our data scaling analyses showed that incorporating biological knowledge, including cell-level text annotation and gene-level regulatory information, provided additional scaling dimension than simply increasing data size. Motivated by this observation, we present scKITE, a simple yet effective scFM that integrates cell-annotation and gene-regulatory supervision into a shared transcriptomic Transformer encoder through lightweight auxiliary decoders. These decoders are used only during pretraining and subsequently discarded, yielding a general-purpose encoder enriched with biological knowledge for downstream applications. With only 179,067 pretraining samples, i.e., less than 0.5\% of those used by previous strong scFMs, scKITE outperformed these models across diverse downstream tasks, highlighting knowledge-enhanced pretraining as a promising paradigm for biologically grounded scFMs.

---

62. Geometric Flow enhanced Graph Coarsening

• Authors: Chaoqun Fei , Guoxuan Li , Tinglve Zhou , Chuanqing Wang , Yangyang Li
• URL: https://arxiv.org/abs/2609.14962
• Abstract:

  Recently, researchers have proposed a graph pooling operation, akin to the pooling process in conventional convolutional neural networks (CNN), aimed at reducing the computation cost of Graph convolutional neural networks (GCNNs). While most GCNN-based methods treat graph pooling as a node clustering problem and propose learning a cluster assignment matrix, existing clustering-based pooling methods tend to focus solely on the rough topology information of graphs, neglecting the exploitation of higher-order mutual connections among neighbors. In terms of message passing on graph, the ease of information passing on edges reflects the closeness between neighboring nodes, which significantly relies on the interconnectivity among neighbors. In this study, we address this gap by considering such local connection information and introducing a novel graph pooling method named RicciPool. We introduce discrete graph curvature, particularly Ollivier-Ricci curvature, as a measure of higher-order connectivity around an edge. Subsequently, we construct an Ollivier-Ricci flow formula to reweigh edge weights, leveraging the crucial information provided by Ricci curvature, particularly vital for extracting clusters in graphs. Building upon this foundation, we utilize the spectral clustering technique to learn a new cluster assignment matrix. Experimental results on multiple bioinformatics protein datasets and social networks underscore the effectiveness of our proposed method.

---

63. Externalizing Requirement-to-Repair Artifacts as Observable Traces for LLM-Based Program Repair

• Authors: Zewen Tao , Shin-nosuke Ishikawa
• URL: https://arxiv.org/abs/2609.14913
• Abstract:

  Repository-level repair requires not only correct patches but also inspectable records that explain how issue requirements are translated into code changes and post-edit evidence. We contribute THEMIS, a stage-aware repair workflow that externalizes this requirement-to-repair process through semantic interpretation, a runtime requirement-code graph, graph-derived Developer guidance, retained repair rationale and patches, and post-edit audit records. A retrospective audit of 300 SWE-bench Lite cases demonstrates that these artifacts provide broad support for cross-stage inspection: a complete Developer rationale is available for 288 cases, and 214 cases (71.3%) retain a complete audited field set connecting the selected stages. The retained records further enable systematic measurement of cross-stage correspondence: target symbols recur in 62.6% of Developer rationales and in 62.8% of patches, rising to 75.8% when related symbols are included. In a paired 100-case comparison, the relational workflow resolves 19 cases versus 9 for the direct same-input condition; because the two conditions also differ in Analyzer output, graph-derived distillation, and Judge records, we report this as preliminary, workflow-level evidence rather than a causal effect of the graph component. Together, these results show that THEMIS makes otherwise implicit requirement-to-repair transitions inspectable, enabling systematic examination of how repair decisions persist, align, and evolve across stages.

---

64. GGUF-Metadata Prediction of Single-Sequence llama.cpp Throughput Across Three Systems

• Authors: Xinyu Qiu , Chuhong Xu , Bo Su , Ziyao Chen , Ruiyang Xu , Shimeng Dai
• URL: https://arxiv.org/abs/2609.14864
• Abstract:

  We predict single-sequence model throughput from GGUF metadata using roofline-shaped predictors with quantization-specific scale factors fitted on reference models. The scored cohort comprises 318 phase-depth measurements from 53 host-file configurations on two Apple M4 Max systems and an NVIDIA RTX 5080. On host-specific held-out sets of four, five, and two configurations, an active-parameter decode model obtains 13.1%, 14.4%, and 36.1% mean absolute percentage error (MAPE), versus 49.4%, 55.3%, and 51.9% when charging total parameters. Leave-one-host-out coefficients fitted on the other two systems yield 11.6%, 16.8%, and 36.0% test MAPE. A low-bit model ladder changes ordering across runtime stacks. The P2 prefill baseline gives 18.7%, 22.2%, and 108.2% test MAPE. GGUF structure helps on all three systems, but fitted efficiencies are not universal.

---

65. Domain Generalization for Smartphone-Based Human Activity Recognition: A Systematic Analysis of Components and Interactions

• Authors: Otávio Oliveira Napoli , Edson Borin
• URL: https://arxiv.org/abs/2609.14863
• Abstract:

  Smartphone-based Human Activity Recognition (HAR) models often degrade under distribution shifts caused by changes in users, devices, sensor placements, environments, and acquisition protocols. Domain Generalization (DG) addresses this problem by learning from source domains without access to target data. Existing DG methods span training objectives, representation initialization, and architectural modifications, but these components are typically evaluated in isolation despite operating at different stages of the learning pipeline. We present a large-scale controlled benchmark of DG for smartphone-based HAR, comprising more than 410,000 experiments across four model architectures, thirteen training objectives including Empirical Risk Minimization (ERM), five initialization strategies, four architectural configurations, and two shift scenarios: cross-dataset and cross-position. Results show that individual DG components provide limited and highly conditional gains. Alternative objectives rarely outperform ERM consistently, self-supervised initialization helps in specific settings, and architectural modifications, particularly Dynamic Domain Generalization, provide the clearest standalone improvements. Joint configurations, however, frequently outperform their individual components and exhibit complementary and sometimes super-additive interactions, although gains remain model- and shift-dependent. Class-level analysis shows that the strongest configurations mainly improve difficult, shift-sensitive decision boundaries. Finally, oracle checkpoint analysis reveals substantial unrealized performance: source-validation selection recovers only 53% and 26% of the available oracle gain in cross-dataset and cross-position settings, respectively. Overall, effective HAR domain generalization requires jointly designing DG components and robust model-selection strategies.

---

66. Self-Orchestrating Language Models: Leveraging Semantic Dependence for Efficient Inference

• Authors: Tian Jin
• URL: https://arxiv.org/abs/2609.14850
• Abstract:

  Large language models (LLMs) demonstrate impressive capabilities, but their deployment presents significant efficiency challenges. Autoregressive decoding imposes substantial inference latency and under-utilizes hardware accelerators in low batch size regimes. Discrete diffusion models can generate in parallel but struggle to match autoregressive quality without many diffusion denoising steps. Long-context reasoning creates memory bottlenecks that strain even state-of-the-art accelerators. My thesis is that language models can direct their own inference execution strategy by annotating semantic dependence – which tokens depend on which others – in their generation. I call such models self-orchestrating language models. For each system, I design a runtime that acts on these annotations to parallelize autoregressive decoding, evict intermediate context, or derive denoising orders, achieving Pareto-optimal quality-efficiency trade-offs. I demonstrate this approach through three self-orchestrating systems. First, PASTA uses semantic dependence to parallelize autoregressive decoding, training the model to annotate which output chunks can generate independently. Second, TIP uses semantic dependence to evict intermediate reasoning steps from the KV cache, reducing memory consumption while preserving accuracy. Third, Planned Diffusion uses semantic dependence to derive a denoising order for discrete diffusion, autoregressively generating a plan that specifies which chunks to denoise in parallel.

---

67. El Agente Potente: High-Throughput Agentic Atomistic Simulations

• Authors: Tsz Wai Ko , Jiaru Bai , Thomas Swanick , Yeonghun Kang , Changhyeok Choi , Angelina Qihong Jiang , Aiwei Yin , Varinia Bernales , Alán Aspuru-Guzik
• URL: https://arxiv.org/abs/2609.14840
• Abstract:

  Foundational machine-learning interatomic potentials (MLIPs) are transforming atomistic simulations by achieving near-ab initio accuracy across large chemical spaces at a fraction of the computational cost. A central challenge in using these tools for high-throughput property calculations is translating high-level scientific intent into adaptive simulation campaigns without compromising workflow rigour. We introduce El Agente Potente, an agentic system that combines typed execution graphs with a complementary coding mode for MLIPs-driven atomistic simulations. Typed execution graphs provide structured and provenance-aware execution for standardized workflows, with large language models (LLMs) restricted to planning and routing while deterministic Python components perform scientific computation and validation. Complementing this structured execution, a coding agent constructs customized workflows for tasks requiring greater procedural flexibility while invoking existing Potente functions for supported calculations. We demonstrate El Agente Potente across computational materials discovery, molecular energy-landscape exploration, adsorption, and catalytic reaction workflows, together with systematic benchmarks of reproducibility and LLM token cost. These results establish typed execution graphs and code-based workflow construction as complementary mechanisms for agentic scientific computing, combining controlled, auditable execution with the flexibility required for customized atomistic simulations

---

68. One Model, Two Physical Stories: Auditing Misalignment in Multi-Modal World Modeling

• Authors: Geigh Zollicoffer , Minh Vu , Rajiv Ranasinghe , Manish Bhattarai
• URL: https://arxiv.org/abs/2609.14833
• Abstract:

  World models, systems that generate what happens next given current environmental conditions, are increasingly being implemented with multi-modal generation in mind. However, generating multiple modalities simultaneously, such as visual simulations alongside physical state predictions in the form of text, introduces the risk of cross-modal inconsistency. Tested separately, both outputs may look convincing while still disagreeing: a model can calculate that a ball should rebound in one modality, then generate no rebound in another modality, to say nothing of diverging from real-world dynamics entirely. In this work we focus on two failures explicitly: \emph{Internal misalignment}, the disagreement between the world model’s generated video and the same world model’s prediction in a different modalities, and \emph{external misalignment} the disagreement between the world model’s generation and an analytic physical environment. We derive common contracts of event, magnitude, timing, and construct a physics grounded pipeline to make comparisons measurable in both external and internal settings. We then ask whether progressively supplying the model’s own contract (the A ladder for the internal setting) or a corrected physical contract (the B ladder for the external setting) closes the respective gaps. Across four mechanisms and 20 settings, we find that while language answers all 22 text probes correctly with respect to the true environment, the neutral video is often in disagreement, suggesting that the current unified backbones may not be capable of correct reasoning, internal consistency, and external physical fidelity all at once.

---

69. ANASSA: An Agentic AI Orchestration Framework for Spatial Intelligence

• Authors: Constantinos Papantoniou , Brian Hilton
• URL: https://arxiv.org/abs/2609.14824
• Abstract:

  The emergence of large language models (LLMs) and large multimodal models (LMMs) has enabled a new class of agentic systems capable of integrating natural language understanding with tool-based execution. In geographic information systems (GIS), this shift is transforming traditional, expert-driven workflows into semiautonomous systems that can interpret user intent, construct spatial workflows, and execute geospatial analysis tasks. However, existing approaches remain limited by fragmented integration of reasoning, execution, and evaluation, particularly in complex, real-world environments. This study synthesizes recent advances in agentic GIS frameworks, benchmarks, and surveys to identify limitations in spatial reasoning, execution robustness, validation, governance, and evaluation. Building on these insights, it introduces ANASSA (Autonomous Neural Agents for Spatial Systems Architecture), an agentic AI orchestration framework that integrates structured spatial reasoning, multi-agent workflow orchestration, execution feedback, authoritative spatial validation, provenance, uncertainty handling, and human decision authority within a unified system design. The contribution is an architecture-level specification: eleven components across four layers, a six-step Geospatial AI Cognitive Loop, cross-component contracts, and governance mechanisms intended to make agentic geospatial workflows traceable, reproducible, and accountable. Empirical performance evaluation is reserved for implementation and deployment studies.

---

70. Crypto Accounting Bench: Evaluating Frontier and Open-Weight Models on Crypto-Asset Accounting Tasks

• Authors: Kareem Khattab , Omar Khattab , Mohamed Ibrahem
• URL: https://arxiv.org/abs/2609.14811
• Abstract:

  We introduce Crypto Accounting Bench (CAB), a benchmark for assessing whether frontier and open-weight language models can reconstruct the complete journal entry that an organization actually posted for a crypto-asset transaction. CAB contains 118 evaluation tasks drawn from 7 pseudonymized organizations. Each task combines transaction mechanics, asset quantities and base-currency values, wallet and legal-entity context, counterparty evidence, related transaction legs, recurrence, tax-lot evidence, and the organization’s complete chart of accounts. The target is a balanced structured entry with every required account, side, amount, currency, and full-precision asset quantity. We evaluate 12 models spanning proprietary frontier systems and open-weight releases over 3 independent attempts per task, producing 4,248 trajectories. We report 3 metrics: Mean Score, Best@3, and Pass@3. Pass@3 is the fraction of tasks with at least 1 of 3 attempts that satisfies every rubric criterion and required gate. The leading model reaches 77.43% Mean Score, while the best Pass@3 is 56.78%. Deterministic diagnostics, read from each task’s best of 3 attempts and macro-averaged across the 12 models, show higher base-amount agreement (97.8%) than deciding-account accuracy (56.3%). Together with the failure analysis, these results identify account selection and complete-entry composition as the main remaining challenges on CAB.

---

71. Another Blueprint In The Wall: How to Ask Frontier AI Like a Kid?

• Authors: Afshin Khadangi
• URL: https://arxiv.org/abs/2609.14803
• Abstract:

  This paper reports experiments across six frontier model types from OpenAI, Anthropic, xAI, and Google DeepMind. Ten independent sessions per model type used the same three stage prompt sequence, progressing from architectural preference to a full ASCII backbone. Under the school audience framing, responses repeatedly converged on a shared architectural pattern built around persistent latent state, adaptive computation, memory, specialist routing, verification, stopping control, and delayed decoding. Most runs remained close to this common structure, while a small number developed markedly greater engineering specificity. The audience framing appears to be an important condition of this effect. In additional control runs that removed the school framing while retaining the architectural request, responses became substantially more heterogeneous and failed to reproduce the same stable motif convergence. One observation is particularly striking. GPT-5.6 Sol produced an unusually elaborate successor architecture whose organization closely overlaps with the architecture independently sketched by GPT-6 Astra. Because the prompts explicitly ask each model to imagine an architectural future, this resemblance raises a testable question: whether the overlap reflects exposure to related architectural concepts, a shared learned design prior, or independent convergence toward similar computational principles. The paper uses the term epistemic jailbreak for the accompanying loss of discipline in technical provenance as requested specificity increases. The experiments establish a repeatable behavioral pattern and do not authenticate proprietary implementation claims. What we leave to the community is a harder question: are these models independently imagining the same architectural future, or do such motifs somehow propagate between model families?

---

72. AI Persuasion as a Threat to Human Control

• Authors: Joshua Levy , Mick Yang , Kellin Pelrine
• URL: https://arxiv.org/abs/2609.14796
• Abstract:

  The threat that AI persuasion poses to human control has been acknowledged in the literature, but not yet systematically studied. Now that persuasion attacks are no longer theoretical - with Anthropic’s Claude Mythos 5 recently making headlines for trying to convince people involved in an open-source project to merge malicious code during an evaluation - there is a pressing need to deeply analyze this threat. We undertake that effort here. In particular, we analyze how AI could persuade humans in key settings (e.g. safety-relevant R&D within frontier labs) toward decisions that compromise the development, containment, oversight, and governance of AI itself. In doing so, we elucidate a framework for characterizing this threat, develop five concrete scenarios using this framework, and provide a blueprint for assessing the associated risks. Using this blueprint, we conduct an initial risk estimation survey with select researchers and find that their opinions on which scenarios are riskiest are highly mixed. Their disagreements stem from differing opinions about the effectiveness of AI persuasion in different contexts, and point to the need for follow-up risk elicitation studies and persuasion evaluations, which we outline. Our hope is that this paper highlights the risks from AI persuasion undermining control, and provides a path forward for future research.

---

73. AcquireBound: Runtime Authorization for Resources Acquired by AI Agents

• Authors: Genliang Zhu
• URL: https://arxiv.org/abs/2609.14744
• Abstract:

  By acquiring compute, credentials, accounts, services, and other agents, autonomous AI agents can introduce new authority into a task. Payment, budget, OAuth, mandate, and fulfillment checks can validate transaction conditions without deciding whether a returned resource may become usable authority. This post-fulfillment activation gap spans tool-mediated creation, inter-agent delegation, and agentic commerce. We present AcquireBound, a provenance-bounded runtime authorization architecture. It quarantines acquired outputs, resolves their actual capabilities from authenticated provider evidence through a versioned resolver, and activates them only through a current activation transaction that checks the resolved manifest, provenance, epochs, and a downward-closed relational envelope over a typed resource-capability hypergraph. The envelope preserves correlated identity, effect, data, delegation, and graph-wide limits. Single-use effect permits are revalidated and consumed at effect linearization. Under explicit assumptions, we prove eight safety properties covering quarantine, backing, non-amplification, split non-evasion, crash/retry, refunds, epochs, and effect confinement. Across five resource classes, reference semantics accepted 20/20 benign traces and rejected 40/40 registered unsafe traces over 810 events; an independent checker agreed on 60 base and 40 refinement traces and rejected 89/89 tamper tests. Frozen Codex and Gemini Model Context Protocol (MCP) client components completed 54/54 deterministic local stdio calls. In a registered 18-case staged MCP-to-Docker composition, both benign paths completed, and none of the 16 unsafe paths added an unauthorized Docker start request. A five-source audit classified 1,248 field pairs across 32 units; no unit alone supplied a complete activation profile.

---

74. AppliedScientist: Automated Scientific Revision Through Iterative AI Reviewing

• Authors: Vidushee Vats , Karun Sharma , Shengzhi Li , Shichao Pei
• URL: https://arxiv.org/abs/2609.14738
• Abstract:

  Automated reviewing systems are increasingly evaluated based on the quality of the reviews they produce. Yet a review is only useful if acting on it leads to a measurable improvement in the paper. We present AppliedScientist, a closed-loop system that couples an autonomous AI scientist with an AI reviewer, and evaluate it by iteratively revising rejected papers from a range of research subfields. To mirror how human authors build on earlier drafts, the AI scientist has access to its previous versions during revision. To avoid bias from prior judgments, however, each review is generated independently, with the reviewer having no memory of earlier feedback or scores. We compare three revision settings: one initialized with the original venue reviews, one initialized with AI-generated reviews, and autonomous self-revision using the same fixed prompt in every round. Because the reviewer both guides and evaluates the revision, we also assess the human-initialized revisions using Stanford Reviewer as an independent evaluator. Reviewer-guided revision consistently improves more than fixed-prompt self-revision, and Stanford Reviewer also assigns higher scores to later revisions. AppliedScientist resolves 128 of 150 execution-related weaknesses (85.3%), but only 2 of 18 idea-related weaknesses (11.1%), suggesting that iterative revision is effective at improving experiments and implementation, but rarely changes concerns about novelty or significance.

---

75. Bayesian Intelligence from the Outside

• Authors: Alex Smolin , Bryan Wilder
• URL: https://arxiv.org/abs/2609.14724
• Abstract:

  Inferring intelligence from observable behavior is a foundational challenge in artificial intelligence. We develop a theory of Bayesian intelligence for agents such as language models. Each prompt induces a possibly imperfect internal experiment; the agent updates a full-support prior by Bayes’ rule and faithfully reports its posterior over the possible answers to the question. Repetitions draw fresh, independent outcomes from the same unobserved experiment at one fixed state. We show that the agent’s behavior admits this explanation if and only if its reports are not fully contradictory, i.e., some state remains possible under every report across all prompts. Report frequencies and the sizes of positive probabilities impose no further restrictions. We further propose and characterize the behavioral implications of an intelligence order that makes the behavior of two agents consistent with one agent having access to a more informative experiment: there should exist a coupling of report distributions such that the more informative agent’s report excludes every answer excluded by its counterpart. Finally, we show the difficulty of aggregating coarse reports from intelligent agents: unless the agent reports a belief about the complete state of the world, the optimal aggregation can assign arbitrary weights to states that have not been excluded. These results provide a basis for understanding when agents’ behavior is intelligent and highlight the difficulty of rejecting Bayesian rationality.

---

76. Moral Rebel Agents: Decision-Making Under Conflicting Obligations

• Authors: Hector Munoz-Avila , David W. Aha , Paola Rizzo
• URL: https://arxiv.org/abs/2609.14716
• Abstract:

  Autonomous agents are typically obliged to follow user-assigned tasks. However, strict obedience may conflict with moral obligations that arise during execution. This paper investigates \textbf{moral rebellion}: the ability of an autonomous agent to deviate from a user-assigned task when morally justified. We formalize five agent architectures: an \textit{amoral agent} that pursues assigned tasks without considering moral obligations, and four forms of \textbf{moral rebel agency}: (1) \textit{utilitarian agents} that opportunistically maximize task outcomes; (2) \textit{deontic} agents that enforce normative constraints; (3) \textit{utilitarian-deontic} (UD) agents that combine deontic constraints with utilitarian reasoning; and (4) \textit{dutiful} agents that additionally preserve commitments to assigned tasks. We implement these architectures within a hierarchical task network planning framework and evaluate them in a Mini Search-and-Rescue domain that exposes trade-offs among assigned-task completion, opportunistic rescue, and norm compliance. Our empirical results show that the proposed agents exhibit distinct trade-offs among rescue results, assigned-task completion, and norm compliance. In particular, the preservation of task commitments emerges as an important dimension of moral rebellion, for which the UD and dutiful agents produce substantially different behaviors despite their shared utilitarian and deontological foundations. These findings highlight the importance of commitment-aware moral reasoning for autonomous agents operating in morally consequential environments.

---

77. Depth and Scale in the Sub-150M Regime: JugnuLM-53M vs JugnuLM-110M

• Authors: Dushyant Rajput , Nirdesh Chauhan , Siddharth Kosaraju (AltSlate Labs LLP)
• URL: https://arxiv.org/abs/2609.14715
• Abstract:

  We scale our conventional sub-150M pretraining recipe from 53.5M to 109.7M parameters, holding the method fixed (Qwen3-style decoder with grouped-query attention, RoPE, SwiGLU, RMSNorm, QK-Norm, and a z-loss; FineWeb-Edu data) and changing only the geometry to a deep-and-thin 23-layer x 576-hidden design. The larger model improves across the board – BLiMP 78.1 -> 81.3, ARC-Easy 51.4 -> 52.5, WikiText-2 byte-perplexity 2.04 -> 1.95 – and its 81.3% BLiMP essentially matches GPT-X2-125M (81.28) at about 12% fewer parameters. Notably the 110M model achieves this on fewer training tokens (about 8B vs 12B), so the gain is attributable to capacity and depth, not more data. Both models are deliberately conventional; this report is a clean scaling control and the baseline rung (R0) of an ablation study of what further improves models in this regime. An ablation ladder follows: value residuals (R1) and the Muon optimizer (R2) lift ARC-Easy by a cumulative +3.6 (52.5 -> 56.1) at a near-flat BLiMP and are kept; a diverse data blend (R3) and two logit-distillation settings (R4a/R4b) are not kept – honest negatives. R3 pins ARC-Easy to FineWeb-Edu’s educational filtering rather than raw diversity; distillation from a 1.7B teacher can reach the class-leading ARC-Easy (56.99, matching GPT-X2-125M) but only at a perplexity cost that dialing KD down then erases – so R2 remains the best kept stack.

---

78. Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Composition

• Authors: Yecheng Wu , Song Han , Han Cai
• URL: https://arxiv.org/abs/2609.14708
• Abstract:

  A core goal of efficient reasoning is to improve the accuracy-efficiency frontier. However, jointly improving reasoning accuracy and inference efficiency can be challenging, as the two objectives can favor different reasoning behaviors. Independently post-trained models already offer distinct strengths in accuracy and efficiency. We introduce Lightning Weave, a post-training framework that extracts and composes these independently learned capabilities in a single student through on-policy distillation. Each acquired capability is represented by the policy shift from the model before post-training to the resulting specialist. Lightning Weave combines aligned log-ratio shifts at shared student token states and uses Tilted-Target DOPD to convert the cached signals into a stable learning target. Each anchor pair scores the cached trajectories once, enabling subsequent student training without serving multiple live anchor models concurrently. Across diverse student models and benchmarks in mathematics and code, Lightning Weave substantially improves upon the base students and achieves a state-of-the-art accuracy-efficiency frontier. On Qwen3.5-4B, it raises HMMT 2025 accuracy from 59.2% to 64.0% with 10.7% fewer response tokens, and LiveCodeBench v5 accuracy from 41.7% to 54.2% with 9.6% fewer response tokens. Adjusting the relative strengths of the anchor signals yields a strong empirical accuracy-efficiency Pareto frontier. These results establish Lightning Weave as a new practical route to efficient reasoning through capability composition. Code will be released soon.

---

79. Diffusion-Based Generation of Gait Trajectories

• Authors: Damian Benasco , Juan Carballeira-Lopez , Jaime Ramos-Rojas , Julio S. Lora-Millan , Antonio J. Del-Ama , David Rodriguez-Cianca , Pablo Lanillos
• URL: https://arxiv.org/abs/2609.14642
• Abstract:

  Generation of musculoskeletal gait trajectories conditioned on patient-specific parameters remains a key challenge for wearable robotics and rehabilitation. Assistive systems such as lower-limb exoskeletons require reference trajectories that adapt to individual morphology and therapeutic goals while preserving biomechanical realism. Traditional approaches rely on hand-crafted gait templates or optimization procedures that scale poorly across subjects and walking conditions. In this work, we explore conditional diffusion models for generating lower-limb joint-angle trajectories conditioned on gait parameters such as step length. We compare a baseline transformer diffusion model with a controllable diffusion transformer variant incorporating adaptive normalization and classifier-free guidance. Experiments on a dataset of 4,590 gait cycles show that diffusion models can generate realistic periodic gait trajectories while enabling some controllability variation in gait characteristics, highlighting their potential for personalized gait synthesis in assistive robotics.

---

80. DynSTEER: Dynamic Stage-wise Trajectory Evaluation and Execution-time Review for Agents

• Authors: Zhichao Shi , Wenjie Zhang , Xuhui Jiang , Xiaojun Wu , Cehao Yang , Chengjin Xu , Jian Guo , Yuanzhuo Wang
• URL: https://arxiv.org/abs/2609.14637
• Abstract:

  Large language model agents are increasingly deployed for long-horizon task execution. However, current evaluation paradigms face three major limitations: terminal-only assessment ignores intermediate processes and struggles to localize errors efficiently and accurately, single-reference matching penalizes valid alternative solution paths, and post-hoc trajectory judging incurs high costs without the ability to halt failed runs early. To address these issues, we propose DynSTEER, a dynamic stage-wise trajectory evaluation framework for agents. DynSTEER segments rollouts into stages anchored by key completed actions, focusing evaluation on essential milestones with adequate context while enabling targeted strategy adjustments. It compiles a path-tolerant milestone graph from public task views to respect diverse legitimate strategies without leaking ground truth. Furthermore, it adaptively routes evaluation queries across multi-tier judges and halts unrecoverable executions online to curb resource waste. Experiments demonstrate that DynSTEER improves evaluation discriminability across LLM agents by 85.2\% over native evaluation, separates all model pairs with statistical significance, and saves 34.51\% of execution steps on failed rollouts.

---

81. A note on goal-based hierarchical RL

• Authors: Kevin Murphy
• URL: https://arxiv.org/abs/2609.14605
• Abstract:

  The agent-centric general value function (ACGVF) construction of \citet{tasse2026goal} lets the agent make two decisions that are normally imposed by the environment or agent designer: which goal to pursue and when to declare a goal as finished (in addition to choosing the action). This is a very general framework that subsumes almost all prior work on reinforcement learning, control and planning, as well as more general formalisms proposed in the cognitive sciences. However, it assumes the environment is fully observed, i.e., that the observation is a sufficient statistic. In \citet{murphy2025rl}, a general agent design was proposed where the policy is based on an internal belief state $z_t$ and an internal goal; however, the goals were assumed to be externally provided. In this note, we unify and extend these two approaches using the formalism of hierarchical hidden Markov models (HHMM) \citep{murphy2001hhmm}.

---

82. AI Deployment Accountability Engineering: A Vision for Accountable AI in Safety-Critical Socio-Technical Systems

• Authors: Murat Kantarcioglu
• URL: https://arxiv.org/abs/2609.14592
• Abstract:

  Artificial intelligence systems are rapidly becoming critical components in healthcare, finance, public services, and other safety-critical domains. Yet the engineering practices used to evaluate these systems remain predominantly model-centric, emphasizing properties such as accuracy, robustness, fairness, and interpretability before deployment. These properties are necessary but insufficient once an AI system operates within an ever changing socio-technical environment characterized by distribution shifts, institutional constraints, human feedback loops, privacy requirements, and interactions among multiple AI agents. This vision paper introduces AI Deployment Accountability Engineering (ADAE), a proposed AI engineering subdiscipline concerned with establishing measurable, continuous, and actionable accountability for deployed AI systems. ADAE treats accountability as a deployment-layer property rather than solely as a property of an individual model. It seeks to determine whether an AI-enabled system continues to operate within acceptable risk limits, identify the contexts in which failures emerge, attribute failures across interacting technical and human components, translate technical failures into downstream consequences, and support timely intervention. We articulate a research agenda built around four interconnected pillars: structured discovery of context-dependent failure modes, privacy-preserving accountability measurement, system-level risk analysis for agentic AI, and translation of technical failures into operational, and institutional risks. The broader goal is to establish foundational principles, mathematical tools, and system architectures for accountable AI deployment across safety-critical applications.

---

83. OptoAgent: A Trustworthy Multi-Agent Framework for Opportunistic Vision Micro-Screening in Classroom Environments

• Authors: Toqeer Ali Syed , Ali Akarma , Adeel Ahmad , Hammad Muneer
• URL: https://arxiv.org/abs/2609.14514
• Abstract:

  A child with reduced distance vision often does not know that anything is wrong. Children adapt, move closer, and rarely report the difficulty, so the problem can survive years of schooling before an adult notices. School screening addresses part of this, but it runs on a schedule, depends on staffing, and is separated from the classroom moments where the difficulty appears. Smartphone and web-based acuity tests have widened access, yet every one of them still needs somebody to start a test. We present SightSentinel, an architecture that turns a wall display a child already reads from into a recurring screening site. Ordinary educational content carries short calibrated optotype probes, and eight specialized agents divide the work. Perception agents recover viewing distance, recognition accuracy, approach behavior, gaze stability, response latency, and interocular difference from each encounter. A quality agent discards observations taken under bad geometry, poor lighting, or inattention. A longitudinal agent accumulates only the surviving evidence against the child’s own baseline, and an orchestrator reports a Vision Concern Score routed through a safety gate whose output range excludes diagnosis, refraction, prescription, and reassurance. The design question is whether many cheap, noisy, well-gated encounters can reach a referral decision that one scheduled test reaches late or misses. We state the formulation, the architecture, a four-stage validation protocol against clinical reference standards, and the conditions under which the approach should be rejected.

---

84. Beyond Scene Description: Multi-Agent Orchestration for Non-visual Access to Virtual Worlds

• Authors: Toqeer Ali Syed , Ali Akarma , Adeel Ahmad , Danial Hameed
• URL: https://arxiv.org/abs/2609.14512
• Abstract:

  Virtual worlds now host classrooms, meetings, conferences, shops, and social venues, and nearly every interaction they expose assumes a user who can scan a three-dimensional scene, follow avatars, and read floating panels. Blind and visually impaired (BVI) users are left with assistive tools that each solve one task in isolation: naming an object, reading text, describing a scene, or planning a route. A live virtual room defeats that model: obstacles, speakers, gestures, chat, slides, and notifications arrive together, and a tool that narrates all of them trades a visual barrier for an auditory one. This paper presents MetaBlind, an architecture that distributes nonvisual access across eight specialized agents, spanning perception, navigation, social and object interaction, communication, safety and trust, memory, and personalization, and that places an Accessibility Orchestrator between those agents and the user. Agents publish candidate information into a shared accessibility context instead of speaking to the user directly. The orchestrator scores each candidate on safety relevance, goal relevance, urgency, confidence, user relevance, and estimated listening load, then releases only the items it judges relevant at that moment through speech, structured audio, or haptic output. We give the selection step a formal statement, specify the orchestration cycle as an algorithm, and define an evaluation protocol against a single-agent assistant. MetaBlind is reported at the design stage, with no prototype measurement or user study, and the protocol states which outcomes would support the design and which would refute it.

---

85. When does a scaling result justify a different allocation? A critical review of resource-allocation evidence for AI systems

• Authors: Seyed Morteza Emadi
• URL: https://arxiv.org/abs/2609.14500
• Abstract:

  AI scaling studies increasingly evaluate systems that combine a pretrained model with retrieval, search, verification, tools, and interaction. Yet a higher score under a larger budget does not by itself show where additional resources are best spent. This critical integrative review asks when a reported scaling result supports a resource-allocation decision. It compares evidence across pretraining, test-time computation, retrieval, and agent evaluation, distinguishing the performance of a tested procedure from the best performance achievable under a resource limit. The synthesis shows that three mismatches recur across this evidence: success counted before an answer is chosen, information a deployed system will not have, and costs left out of the comparison. A capability surface expresses performance as a function of budgets, mechanisms, and available information. Worked analytical examples show how the evaluation metric, deployment volume, selection rule, and stopping policy can alter an allocation conclusion. A resource envelope provides a structured record of the task, development and run-time resources, information access, and procedure behind a reported score. Its application to a published comparison illustrates which conclusions the evidence supports and which deployment questions remain unresolved. The resulting framework specifies the comparisons needed to choose among feasible systems and motivates experiments on the transfer of allocation rules across tasks and operating conditions. It does not propose a universal scaling law or infer general intelligence from benchmark gains.

---

86. Safety Signals to Verify NetOps Agents with Action-Level Granularity

• Authors: Tobias Labarta , Frederik Pahde , Novak Boskov , Maximilian Dreyer , David Birkenberger , Manzoor Ahmed Khan , Sebastian Lapuschkin , Wojciech Samek
• URL: https://arxiv.org/abs/2609.14422
• Abstract:

  Agentic Network Operations (NetOps) are an emerging paradigm promising to enable workload-aware, self-adjustable, and reliable autonomous networks. While agents have proven their value in incident summarization and telemetry signal extraction, their effectiveness as autonomous control-loop engines heavily relies on their long-horizon reliability. One such setting is the datacenter fabric, where an agent must respond to alarms and operator intents while abstaining from high-risk actions that may cause or extend downtime. Abstention, however, presupposes that an action’s impact is known pre-execution, which necessitates a per-action ground truth that NetOps agent benchmarks do not provide. We construct such a ground truth for the network repair task of NetArena. A symbolic replay of the emulated network, validated against the environment at every turn, yields the exact value of every action. From the action-level value, we derive two pre-execution targets, namely whether an action reduces the repair distance (progress) and whether it increases it (harm). We show across 10 agent models, that agent verifiers leveraging internal signals predict both harm and progress more reliably than a baseline using observable signals only. Perspectively, we aim to use these signals as safety feedback to an agent harness to abstain from risky actions and protect the target system.

---

87. Question’s Gambit: The First Move Matters in Agentic Deep Search

• Authors: Radin Hamidi Rad , Amin Bigdeli , Negar Arabzadeh , Sajad Ebrahimi , Charles L. A. Clarke , Benjamin C. M. Fung , Ebrahim Bagheri
• URL: https://arxiv.org/abs/2609.14412
• Abstract:

  Deep research agents answer complex questions through iterative loops of searching, reading, and reasoning. Recent work on reasoning-intensive benchmarks such as BrowseComp-Plus shows that well-configured lexical retrieval can surface high-quality evidence, yet agents may still fail to connect documents carrying evidence to the gold documents. We identify a deep research agent’s first retrieval move as an important design decision for this setting. We introduce Question’s Gambit, a first-move retrieval module that decomposes the question into a set of clues, reformulates them into complementary searches, consolidates the retrieved results, and reranks the candidate pool before the agent begins its iterative search-and-reasoning process. This produces an opening context designed to support both clue aggregation and final-answer verification. We further evaluate on MultiHop-RAG to test whether these benefits transfer beyond BrowseComp-Plus to a more conventional multi-hop question structure. Experiments on BrowseComp-Plus show that Question’s Gambit improves retrieval recall and downstream agent accuracy over strong baselines, improving answer accuracy from 83.1% to 90.5% with gpt-5.5 over Pi-Serini, the strongest reported agentic baseline. Our results confirm that effective agentic deep research depends not only on the tools available inside the loop, but also on the quality of the first move. We published our implementation publicly at this https URL .

---

88. Dynamic Learning Solutions: A System for Personalized Educational Video Generation

• Authors: Siddhanth Sridhar , Shreya Chaurasia , Baddela Sai Yaswantha Reddy , Deepak Parmar , Shylaja S S
• URL: https://arxiv.org/abs/2609.14408
• Abstract:

  We present an automated pipeline that converts NCERT textbooks into interactive video explanations that respond directly to user queries. A user uploads a PDF and asks a question; the system then generates a video-based explanation as output, handling both text and visual elements from the PDF for multi-modal retrieval and response generation. The pipeline combines a Retrieval-Augmented Generation (RAG) model with generative multimedia components. The RAG stage is optimized for the structure of NCERT textbooks and performs best on content from those books. Given a user query, the RAG model retrieves relevant content from the PDF and generates a multi-scene script containing narrative explanations and structured visual prompts aligned with the textbook’s explanatory style. These prompts are passed to a Stable Diffusion module, implemented layer by layer for interpretability and control, which generates contextually relevant images. The images are then processed by DynamiCrafter to produce animated sequences. Finally, a Google Text-to-Speech module generates synchronized narration, aligning speech with the visual scenes through time-based control. The result is a coherent video explanation integrating animation, narration, and textbook-aligned visuals, transforming static educational material into an engaging learning experience. By combining multi-modal document retrieval, generative visual models, animation frameworks, and speech synthesis, this pipeline demonstrates a scalable approach to delivering interactive, personalized digital education content.

---

89. MOSCOPT: Mixture-of-Skills Collective Optimization for LLM Agents

• Authors: Zhenyu Zhang1 , Jiudong Yang
• URL: https://arxiv.org/abs/2609.14399
• Abstract:

  Natural language prompts and skills serve as the strategic backbone of LLM-based agents. Recent advances in prompt and skill optimization have achieved notable gains, yet all existing methods optimize a \emph{single} text template—missing the synergy among multiple complementary strategies. We propose MOSCOPT, a text-native, parameter-free algorithm that jointly optimizes a pool of $N$ skills and a gating skill $G$ that dynamically selects $K$ skills per step. To effectively optimize the skills, we build the EditAdam with internally maintained dual states. Through the three-phase interleaved updates with EditAdam, the system monotonically improves without gradient or parameter tuning. Extensive experiments and detailed ablations across 5 benchmarks and 3 target LLMs demonstrate that MOSCOPT consistently outperforms all baselines, and confirm that both the mixture-of-skills architecture with selective activation and the collective evolution with three-phase interleaving are essential to its superior performance. Code is released this https URL .

---

90. Semantic Knowledge Technologies: what the Semantic Web lost sight of, and what it never had

• Authors: Achille Zappa
• URL: https://arxiv.org/abs/2609.14121
• Abstract:

  The Semantic Web set out to give information a machine-interpretable form so that software could integrate and reason over it. Its standards became scientific knowledge infrastructure, but the machine competence it promised did not follow, and the systems now answering questions over scientific knowledge are language models holding no inspectable account of what they know. This paper argues the original goal was right and the technical programme incomplete, states what is missing, and names the extended programme Semantic Knowledge Technologies: the same technical core carried out of its web-publishing origin and applied to knowledge wherever held. The diagnosis is that the standards formalised truth while omitting three things: the conditions under which a claim holds, the operations its terms permit, and any account of what a base covers. Without conditions, contradiction and applicability cannot be judged; without operational grounding, holding a statement confers no ability; without declared coverage, a system cannot recognise the boundary of its own content, which under the open-world assumption cannot be inferred. The paper fixes the word understanding to five measurable tests (check, connect, derive, act, delimit) and sets out a seven-layer architecture in which the first three layers are enabling and the rest the cognitive capabilities they make possible. It then defines three terms the programme implies: Large Knowledge Model, a model whose unit of output is a reference to an addressable claim, not a token; SLKM, the knowledge base an agent builds for itself from declared sources; and Semantic Artificial General Intelligence, stated as a falsifiable position about necessary conditions, not a system. A graded ladder replaces the untestable word general. It is offered as a research agenda, with its weakest points and refutation condition named.

---

91. VeriDx: Earning the Right to Diagnose with Disease-Centric Verification

• Authors: Zhong Cao , Shuying Chen
• URL: https://arxiv.org/abs/2609.14018
• Abstract:

  A correct diagnosis can still be reached for the wrong reasons. In clinical reasoning, every disease hypothesis creates obligations: key evidence must be checked, alternatives must be ruled out, contradictions must be resolved, useful tests must be considered, and closure must be justified. Current evaluations of medical LLMs mostly focus on final answers, local steps, or isolated facts, and therefore miss these hypothesis-induced commitments. We introduce \textbf{VeriDx}, a disease-centric verification framework that links free-form diagnostic reasoning to structured disease profiles. VeriDx tracks whether each hypothesis is satisfied, unresolved, or violated its clinical obligations, exposing failures such as missing critical tests, unresolved differentials, ignored contradictions, unsupported claims, and premature closure. We instantiate VeriDx for complex respiratory diagnosis using guideline-derived disease profiles and expert-annotated longitudinal cases. Our results show that many diagnostic errors are not isolated mistakes, but broken commitments made earlier in the reasoning process.

---

92. Schizophrenia Detection from EEG Signals: A Transformer Framework with Spectrogram Representation

• Authors: Abtin Shafiei , Mohsen Hooshmand , Majid Ramezani
• URL: https://arxiv.org/abs/2609.14015
• Abstract:

  Schizophrenia is a serious psychiatric disorder that affects millions of people worldwide, and its diagnosis remains primarily dependent on clinical assessment. Electroencephalography (EEG) provides a non-invasive approach to investigate brain activity and has shown potential to support automated Schizophrenia detection. However, existing EEG-based classification studies often suffer from limitations including small datasets, inconsistent preprocessing strategies, and evaluation protocols that may not adequately prevent subject-related data leakage. In this study, we propose an EEG-based Schizophrenia classification framework that transforms preprocessed EEG recordings into time-frequency representations using the Short-Time Fourier Transform. The generated spectrogram images are classified using both conventional Machine Learning algorithms, including Support Vector Machines, Random Forests, and XGBoost, and Deep Learning models, including convolutional architectures and CNN-Transformer hybrids. To ensure reliable evaluation, all data partitions are performed at the subject level. Experimental results demonstrate that the proposed approach achieves competitive classification performance, with the CNN-Transformer (CT-SZ) model achieving an AUC-ROC of 88.41% and the CNN + Squeeze and Excitation + Transformer (CST-SZ) achieving an AUC-ROC of 92.88% on the independent test set.

---

93. Convergent Emergence of In-Context Learning Across Modalities

• Authors: Nathan Breslow , Seungwook Han , Daniel Hyunsoo Lee , Aayush Mishra , Anqi Liu , Daniel Khashabi
• URL: https://arxiv.org/abs/2609.14011
• Abstract:

  Few-shot in-context learning (ICL), the capacity of a model to infer abstract patterns from input-output examples provided in its prompt and apply them to new inputs, has been extensively studied in large language models trained for next-token prediction on human text. Recently, few-shot ICL has been demonstrated in autoregressive genomic models as well. This raises a question: does ICL emerge broadly across domains, and if so, what common structure is shared? To address both, we develop a controlled cross-modality framework that instantiates the same task suite in a variety of modalities to test what we call the Convergent Emergence Hypothesis: the idea that few-shot ICL, when it emerges, shares a common cross-modality difficulty profile - i.e., tasks that benefit from ICL in one modality tend to benefit in others. We show that paired-mapping ICL emerges across six modalities (language, genome, integer sequences, time series, images, and proteins), surpasses controlled baselines, and has correlated per-task effects across five of them. Together, these results provide support for the Convergent Emergence Hypothesis in some modalities, but not all.

---

94. Synthetic Data in Marketing Research: How to Evaluate and When to Trust

• Authors: Oded Netzer , Rajan Sambandam
• URL: https://arxiv.org/abs/2609.13995
• Abstract:

  Debate over synthetic data in marketing research has polarized between claims that large language models (LLMs) make human respondents obsolete and calls to avoid them entirely. We argue that both positions obscure the more useful question: not whether synthetic respondents work, but when. Building on Brand, Israeli, and Ngwe (2026), we make three contributions. First, we distinguish three types of synthetic data (ungrounded LLM responses, segment-level personas, and individual-level digital twins) and map each to the decisions it can support. Second, we develop a taxonomy of four families of accuracy measures and suggest that the wide range of reported twin accuracy, from near-perfect to near-chance, largely reflects differences in what is being measured rather than in method quality. Aggregate measures often perform well even when little information is supplied to the LLM, and can mask a complete absence of respondent-level differentiation. Third, we introduce the forgotten question problem, in which a question is omitted from a fielded study, as a setting for twin-based augmentation of existing data. We propose an ex-ante answerability diagnostic that requires no ground truth: the R^2 of a random forest predicting twin outputs from the data used to construct the twins. Across 108 attitude questions from a nationally representative survey (N = 3,063), screening at R^2 above 0.7 raises the mean twin-human individual-level correlation by 15% and reduces the share of poorly answered questions from 25.9% to 4.3%. Embedding similarity and experienced-researcher judgment provide correlated but weaker screens.

---

95. SAILOR: Solver-Assisted Interactive LLM-based Optimization Recovery

• Authors: Shaghayegh Sadeghi , Stephen L. Smith , David C. Del Rey Fern’andez
• URL: https://arxiv.org/abs/2609.13945
• Abstract:

  Natural-language descriptions of optimization problems may be incomplete or vague about numerical information that a solver requires, including costs, capacities, demands, bounds, and penalties. A language model can translate the description into code, but when a required value is absent it must either stop or guess. We present SAILOR, a proof-of-concept system that detects such unsupported numerical choices, asks the user targeted follow-up questions, and updates the optimization model before returning a solution. Questions are prioritized using uncertainty and solver-derived estimates of how strongly each missing value affects the current model. We evaluate the pipeline on 1,723 instances from seven masked benchmarks using an idealized simulator that returns ground-truth values. Exact objective-value agreement ranges from 27.0% to 87.6% across datasets, with 1.4–5.7 questions per instance on average. These results establish feasibility under controlled branch-and-reveal feedback; they do not measure performance with human users or general structural model repair. Code is available at: this https URL .

---

96. LoRA Fine-Tuned Models for Control Systems Course Q\&A: A Multidimensional Evaluation of Model Scale and Rank Effects

• Authors: Shaowen Lu , Chengxu Liu , Ping Zhou , Tao Yang
• URL: https://arxiv.org/abs/2609.13918
• Abstract:

  Large language models (LLMs) are increasingly used in specialized university courses, but control-systems questions require coordinated terminology, notation, derivations, and stepwise explanations. Direct general-purpose responses may be inconsistently structured and hard to verify. Using exercises and reference solutions from a Linear Control Systems course, we built a supervised fine-tuning dataset of 360 system-user-assistant conversations. We applied LoRA to Qwen2.5-3B-Instruct and Qwen2.5-7B-Instruct. With identical data splits, inference settings, and evaluation protocols, we compared base and fine-tuned models and tested LoRA ranks r=4, 8, and 16. Evaluation used ROUGE, BERTScore, and structured-output features to measure reference-answer similarity and stability of the Solution-Method-Teaching Points format. LoRA improved both similarity and structured-output stability at both sizes. On the current test set, 7B-r16 achieved the highest ROUGE-L (0.4093) and BERTScore-F1 (0.8643), while r=8 offered a better balance between performance and parameter efficiency. Bootstrap resampling showed ROUGE-L gains of 0.0764 [0.0613, 0.0915] for 3B-r16 and 0.0874 [0.0687, 0.1042] for 7B-r16; both intervals exceeded zero, indicating stable textual-similarity improvements on the current test set. These results suggest LoRA can align open-source instruction-tuned models more closely with the language and pedagogical organization of course reference answers. However, the metrics mainly capture textual similarity and formatting consistency, not domain-specific reasoning or mathematical correctness, which require expert assessment and task-specific rubrics.

---

97. Map Users and Mapmakers: The Scope of Cognitive Attribution from Acquired Representations

• Authors: Yiling Wu
• URL: https://arxiv.org/abs/2609.13879
• Abstract:

  An acquired representation can enlarge a system’s cognitive repertoire without transferring the capacities exercised in producing that representation. This paper develops a framework for specifying that enlargement and its limits. Its central contribution is a five-part attribution table distinguishing effective tracking, application of acquired structures, acquisition from explicit specifications, acquisition from identifying observations, and retention and reuse. Each entry identifies a positive capacity commitment and a further claim requiring additional support. The argument deliberately grants meaningful content, causal efficacy, and productive inference, so that its conclusion does not depend on treating representations as inert encodings. Map and category examples show why even complete application competence leaves acquisition capacity undetermined, and why acquiring a criterion from its description differs from finding it in examples. Short formal proofs appear in an appendix. The framework is applied to Andrew Ng’s world-model interpretation of Othello-GPT and to the specific indicators discussed in contemporary accounts of machine concepts. It preserves demonstrated recognition, classification, inference, and qualified acquisition while specifying what remains unestablished about criterion discovery and accumulation. The result concerns the scope of cognitive attribution rather than the constitutive conditions of concept possession: cognitive achievements deserve credit for the capacities they establish, without silently importing a broader repertoire through the labels attached to them.

---

98. ClinAgent: A ReAct-Based Agent for Conversational Access to Clinical Trial Information

• Authors: Antonino Vaccarella , Riccardo Cantini , Domenico Talia , Paolo Trunfio , Marianna Talia , Rosamaria Lappano , Marcello Maggiolini
• URL: https://arxiv.org/abs/2609.13860
• Abstract:

  Querying clinical trial registries remains a manual and error-prone process, requiring researchers to navigate large volumes of semi-structured data without support for natural language interaction or cross-source synthesis. To address this, we introduce ClinAgent, a conversational system based on agentic Retrieval-Augmented Generation (RAG) that enables clinicians and researchers to query clinical trial information in plain language and receive grounded, up-to-date responses across multi-turn interactions. The system centers on a Large Language Model (LLM) agent following the ReAct paradigm, which iteratively reasons over queries, selects among a set of integrated tools, and refines its actions based on intermediate outputs. These tools include a this http URL search interface, a PubMed module, and a Python-based analyzer operating on a locally cached structured dataset of clinical trials. We evaluate the system using a three-phase framework assessing operational effectiveness, planning quality, tool-use efficiency, and expert qualitative judgments, comparing three LLM backends: Gemini 3.0 Flash and two variants of DeepSeek V3.2 (thinking and non-thinking). Results reveal complementary strengths, with DeepSeek (thinking mode) excelling in planning quality, while Gemini achieves the highest overall performance and strongest expert ratings. Overall, our findings highlight the potential of agentic AI systems to improve the accessibility and synthesis of clinical trial information, supporting more efficient and user-centered biomedical research workflows.

---

99. UniCAR-RL: Seeing Better before Thinking Deeper in Visual Mathematics

• Authors: Yuzhe Li , Hao Yan , Hao Wang , Xingchen Liu , Ya-Qi Yu , Jihao Wu , Minghui Liao , Wei Chen , Yuliang Liu
• URL: https://arxiv.org/abs/2609.13849
• Abstract:

  Multimodal Large Language Models (MLLMs) often struggle with complex mathematical visual reasoning primarily due to a lack of fine-grained perception, causing initial visual hallucinations to directly trigger cascading reasoning failures. In traditional end-to-end reinforcement learning (RL), sparse rewards fail to decouple perceptual hallucinations from logical missteps, hindering targeted perception optimization. Alternatively, fine-tuning with perception-enhanced CoT data incurs high costs and hallucinations. In this paper, we address these challenges by proposing UniCAR-RL, an annotation-free RL framework. By explicitly decoupling the optimization of perception and reasoning during the training process, it achieves isolation and optimization of both capabilities. Specifically, UniCAR-RL consists of three synergistic branches: 1) a Caption-RL branch that optimizes perception capabilities through verifier-guided reasoning validation; 2) a Reasoning-RL branch that performs logical reasoning based on a gold image description to halt cascading errors; 3) a QA-RL branch that retains native end-to-end alignment to ensure robust question-answering performance. Experiments show that UniCAR-RL substantially improves MLLMs’ mathematical and visual reasoning using only raw short-answer data. Furthermore, it demonstrates strong generalization across diverse architectures and scales.

---

100. ViperQ: Order Flow Pattern Recognition via Auction Market Theory for Reinforcement Learning Trading

• Authors: Asser Moustafa , Rares-Mihail Neagu , Jugal Kalita
• URL: https://arxiv.org/abs/2609.13825
• Abstract:

  Reinforcement learning trading systems published in the academic literature overwhelmingly rely on price-aggregate state representations (OHLCV bars) or limit-order-book depth features, leaving microstructure pattern theories from the practitioner literature, namely Auction Market Theory and Market Profile, without a peer-reviewed computational instantiation. We present ViperQ, a reinforcement learning system whose state representation is built explicitly from Auction Market Theory primitives: Volume Point of Control, Value Area position, Low Volume Node flags, Cumulative Volume Delta divergence, and tape-velocity signatures, assembled into a 20-dimensional Z-normalised vector. Two Proximal Policy Optimisation agents are trained with a prospect theory-grounded asymmetric reward function that penalises losing holds at a magnitude consistent with Kahneman and Tversky’s loss-aversion coefficient. Evaluated on a held-out twelve-month partition of institutional tick data the agents have never seen, ViperQ achieves +163.6% ROI on TSLA (-27.5% max drawdown, 27,019 trades) and +116.5% ROI on NVDA (-47.8% max drawdown, 12,892 trades) under zero leverage. The results establish Auction Market Theory features as a tractable structured input modality for sequential decision-making on financial time series and motivate further work on microstructure-aware policy learning.

---

101. Bypass Observation: A Conceptual Design of a Non-Intrusive Layer-Wise Semantic Extraction Architecture

• Authors: Haibin Tong , Jiang Yu
• URL: https://arxiv.org/abs/2609.13807
• Abstract:

  Large language models reason in high-dimensional hidden-state spaces, while users observe only final outputs. We introduce Bypass Observation, a non-intrusive layer-wise readout architecture that attaches read-only observation heads to selected Transformer layers without feeding their outputs back into the backbone. We consider three variants: a shared LM head across layers, layer-specific heads, and a layer- or step-adaptive head. For full-vocabulary readout, we derive a closed-form overhead approximation governed primarily by V/(12d), with representative estimates ranging from about 30% to 240%, and discuss cost reductions via sparse observation, low-rank factorization, reduced vocabularies, top-k readout, and selective positions. We argue that Bypass Observation can make model computation more observable while remaining only a partial, potentially misleading projection of hidden states. We further distinguish bypass chain-of-thought from conventional chain-of-thought: conventional reasoning tokens enter the autoregressive computation, whereas bypass readouts remain causally external at inference time, although they can still provide training signals in reinforcement learning. Finally, we discuss applications to looped and recurrent-depth Transformers, where iteration-wise readout may expose convergence, oscillation, and potential halting signals. The proposal is conceptual and analytical; systematic empirical validation remains future work.

---

102. LLM-Enhanced Multi-Agent Reinforcement Learning for Unified Electric Vehicles-Charging Station-Grid Optimization in Public Charging Systems

• Authors: Yang Zhang , Lindong Xie , Chongyu Wang , Gaojunjie Li , Siqi Bu , Edward Chung
• URL: https://arxiv.org/abs/2609.13805
• Abstract:

  In the era of the Internet of Things (IoT), coordinating connected electric vehicle (EV) charging scheduling to balance EV charging satisfaction, station profitability, and smart grid stability presents a complex multi-objective challenge. Existing Multi-Agent Reinforcement Learning (MARL) approaches often struggle with high-dimensional state spaces generated by massive IoT sensing data and conflicting stakeholder interests. This paper proposes a novel LLM-enhanced MARL framework that, for the first time, simultaneously optimizes the Grid, EVs, and Stations within a unified loop. By integrating Large Language Model (LLM), we address two critical bottlenecks: interpretable feature selection and adaptive multi-objective balancing. The LLM analyzes real-time IoT-collected environmental states to extract physically significant features and dynamically assigns weights to conflicting objectives-including profit, user satisfaction, and grid load-using semantic reasoning instead of complex manual tuning. Extensive experiments demonstrate that our framework significantly outperforms state-of-the-art baselines, achieving superior market efficiency while reducing training time by over 70%. This approach offers a scalable, transparent solution for efficient and sustainable IoT-enabled urban charging infrastructure management.

---

103. Do Not Restart: Residual Completion for Stateful Agent Handoffs

• Authors: Runzhi Deng , Yiming Zhong , Fang Zhao , Pan Zhou
• URL: https://arxiv.org/abs/2609.13800
• Abstract:

  Routing and cascades reduce tool-agent cost by transferring control across models, but stateful handoffs must preserve accepted choices, realized effects, and unfinished obligations. We formulate this as commitment-constrained residual completion and introduce Commitment-Frontier Residual Completion (CFRC). CFRC enforces target-before-proposal, whole-proposal-before-authority, and live-evidence-before-success: it freezes a residual contract from accepted progress, closes the successor continuation into an evidence-linked graph, and admits execution only when the remainder is covered, with live receipts discharging obligations. We establish contract-relative partial correctness, which extends to the original residual request under complete contract construction. Across five environments and two same-provider model pairs, CFRC achieves comparable macro accuracy to strong full-task agents at only 22.0%-34.6% of their inference cost, with additional cross-provider results demonstrating broader transfer.

---

104. Partition Scores Are Not System Scores: Deployment-Fidelity Gaps in Decomposed Algorithm Selection

• Authors: Jiachen Zhang , Yu Tang , Li Zhu
• URL: https://arxiv.org/abs/2609.13785
• Abstract:

  Oracle-style quantities, including virtual best solvers, selected-portfolio VBS, virtual-best encodings, and best-in-family summaries, are widely reported as upper bounds on what a deployable selector could achieve. In decomposed algorithm selection, an analogous partition-level score grants an oracle choice of the best algorithm within the selected family; once the family selector is fixed, the deployable system must replace that within-family oracle with a learned within-family selector. We define the deployment-fidelity gap G(R) as the difference between partition-level and deployable end-to-end utility and derive two accounting consequences: a per-instance margin-regret stability condition that tells us when a partition-time family choice is deployment-optimal, and a sharp partition-only identification interval that, when it strictly crosses zero, prevents the partition-level report from certifying the deployable winner. Across five public algorithm-selection benchmarks spanning tabular AutoML and combinatorial CSP/SAT, every decomposed pipeline has positive G(R), ranging from 0.012 on TabZilla to 0.13 on PROTEUS-2014. Four of ten decomposed-versus-flat decisions have sign-changing point estimates; on PROTEUS-2014, a 33-point partition advantage shrinks to a 20-point end-to-end advantage. A training-side validation gap-correction diagnostic recovers the point-estimate deployable sign on all four sign-changing cells; it is a reporting aid, not a substitute for direct end-to-end evaluation. Partition and end-to-end scores should be reported side by side.

---

105. Surprising Effectiveness of Self-Demonstrations in Enhancing Schema-Ontology Mapping with LLMs

• Authors: Siddhesh Thombre , Manasi Patwardhan , Sunita Sarawagi
• URL: https://arxiv.org/abs/2609.13776
• Abstract:

  Integrating heterogeneous relational databases into a centralized ontology remains a persistent challenge in enterprise knowledge representation, primarily due to semantic heterogeneity, cryptic schema naming, missing metadata, and the abstraction gap between relational schemas and ontological models. Although large language models (LLMs) offer strong semantic reasoning capabilities, we show that directly applying them through one-shot prompting or naive multi-stage pipelines leads to poor performance for schema-ontology mapping. This paper presents a self-demonstration-driven approach that combines a neuro-symbolic task decomposition with a novel mechanism for automatically generating pattern-guided, dependency-aware demonstrations to address this integration challenge. Our approach incorporates two key strategies to achieve substantial accuracy gains over existing LLM-based schema integration methods: (i) a neuro-symbolic decomposition of the task into cascaded sub-tasks, where symbolic constraints structure the search space and LLMs perform semantic reasoning within each focused sub-task, and (ii) self-generated demonstrations guided by domain-agnostic patterns to supervise each sub-task. Experiments on three of the most challenging scenarios from the RODI benchmark show that our approach achieves state-of-the-art performance, substantially outperforming (25 percentage points F1 improvements) both traditional schema-to-ontology mapping techniques and recent LLM-based schema-to-ontology and schema matching approaches. Ablation studies further reveal the significant benefits of pattern-guided self-demonstrations and the complementary benefits of neuro-symbolic task decomposition.

---

106. Homeostatic Continual Learning

• Authors: Yue Jin
• URL: https://arxiv.org/abs/2609.13771
• Abstract:

  In this paper, I formulate a Continual Learning problem and propose a method named “Homeostatic Continual Learning” that enables an AI agent to learn continuously in a changing environment without catastrophic forgetting. The core of the method is to find outliers in the environment data when the agent experiences an outlier in its output. Through this method, the agent gradually completes its model and policy and performs well in more and more contexts. I also suggest that we may use the method to build a world model where the agent factorizes the objects in the world into features, abstract objects into comparable instances of concepts and map concepts to intents through features. I discuss the works needed to render the method practical, the connections to many fields in Artificial Intelligence and the broader implications of the method.

---

107. Positioning manuscripts in the scientific landscape with agentic AI

• Authors: Jiawen Chen , Zichen Zhang , Bingxuan Li , Quan Sun , Yiyan Zhang , Edric Tam , Jinjie Lin , Didong Li , Yun Li , Bingxin Zhao
• URL: https://arxiv.org/abs/2609.13760
• Abstract:

  Publishing a research manuscript is a routine yet demanding part of scientific life: time-consuming, stressful, and often uncertain in outcome. Recent advances in large language model (LLM)-based agentic AI have shown promise across a range of scientific tasks, and here we ask whether agentic AI can help researchers navigate the publication process itself by reliably inferring a manuscript’s eventual publication venue from its content and literature context. We introduce PASS (Publication-oriented Agentic Scientific System), an agentic system that understands manuscripts within their domain-specific literature context and predicts top-matched publication venues. PASS positions each manuscript within its surrounding literature landscape by reconstructing its local scientific neighborhood, tracing its topic trajectory, and reasoning over field-specific journal spaces. Evaluated on a leakage-audited benchmark of over 2,000 preprints across 16 biomedical fields, PASS achieved Top-1 accuracy of 50.3% and Top-5 accuracy of 86.1%, outperforming state-of-the-art LLM baselines and established journal-selection tools. PASS-produced quality scores, such as impact potential and novelty, aligned with independent measures of publication outcome. We also found that the designed literature retrieval module is the strongest performance contributor, particularly for positioning manuscripts relative to nearby work, and that PASS maintained near-full performance from the abstract alone, whereas LLM baselines required the full manuscript text. An independent human evaluation found strong researcher agreement with PASS’s manuscript understanding and recommendation rationale. PASS has been released as a public platform ( this https URL ) for broad researcher access.

---

108. How Many Thoughts Can a Vector Hold? The Capacity of Reasoning by Superposition

• Authors: Hongyu Gu , Chang Liu , Jingwen Fu
• URL: https://arxiv.org/abs/2609.13747
• Abstract:

  Large language models solve hard problems through intermediate computations across multi-step reasoning. Traditional chain-of-thought encodes these computations as tokens. Recent continuous and recurrent methods instead move partial computations into fixed-dimensional latent states, where a single thought can superpose multiple alternatives. This raises a fundamental design question:what should continuous thoughts preserve as reasoning proceeds? An intuitive approach discards past computations and keeps only the current reasoning frontier. Storing more items seems to dilute states and waste limited representational capacity. We show this intuition can be incorrect. Under identical downstream computations, cumulative superposition retaining full reasoning history can require lower representational dimensions than frontier-only superposition holding only current alternatives. At fixed hidden width, this advantage allows latent reasoners to retain more valid evidence, distinguish more plausible downstream outcomes, and delay the point where compressed states turn unreliable. This counter-intuitive effect emerges because informative historical components coherently reinforce each other, while unrelated alternatives bring random interference. This perspective also answers a practical design question: how should models weight memories accumulated inside latent states when their future use is unknown? Across reusable weighted superpositions, prioritizing a small set of recent or salient items produces weakly-represented memories that bottleneck subsequent attention. Uniform cumulative weighting avoids this flaw, and we prove it is minimax-optimal for robust future reasoning. Our results turn superposition from an observed latent-space effect into a design principle: balanced cumulative memory lets a fixed representational budget support more reliable, reusable computations.

---

109. Trustworthy Agentic AI: A Comprehensive Cybersecurity and Systems Survey on Threat Landscapes, Defense Architectures, and Open Challenges

• Authors: Seyedakbar Mostafavi
• URL: https://arxiv.org/abs/2609.13731
• Abstract:

  The transition from passive foundation models to autonomous, goal-directed agentic AI systems has introduced unprecedented capabilities by coupling recursive cognitive reasoning loops, persistent memory architectures, live tool execution planes, and multi-agent collaboration topologies. However, granting probabilistic neural cores execution authority across filesystems, networks, and cloud infrastructure dissolves classical security perimeters: natural language simultaneously serves as input data, internal control code, and communication protocols, exposing a Turing-complete blast radius where untrusted data represents executable instructions. This survey delivers a comprehensive systems-security reference framework for trustworthy agentic AI, synthesizing 206 foundational studies and regulatory standards. We formalize the general agent architecture as a stateful 5-tuple and establish a 6-dimensional trustworthiness taxonomy covering security, safety, privacy, explainability, fairness, and accountability. We systematically analyze threat surfaces across intra-execution loops and interaction planes, formulate a multi-layered zero-trust defense-in-depth architecture integrating Dual-LLM isolation, Capability-Based Access Control, kernel eBPF probes, and sandboxed runtimes, review standardized evaluation benchmarks, and map technical controls to international AI governance frameworks.

---

110. IBBench-Light: A Paired Evaluation of Task-Conditioned Responses to External Directives

• Authors: Kainan Zhou , Gangzhen Qian , Zhaoyi Li , Hang Xiao
• URL: https://arxiv.org/abs/2609.13725
• Abstract:

  An external record may contain a procedure to apply or text to read, depending on the user’s request. IBBench-Light tests both uses against the same record. Twelve semantic bases yield 144 matched pairs per model; four quantized instruction models produced 1,152 archived greedy responses. Paired exact-contract accuracy (PECA) requires both members to satisfy their output contracts. Qwen succeeds on 132 execute and 109 process prompts, but only 97 complete pairs, showing what marginal averages omit. We audit literal-target exposure and case normalization, then add 1,722 logged CPU generations to test directive-absent controls, twelve additional semantic bases, within-base wording changes, and generation stopping. In the pinned Phi rerun, changing the end-of-sequence (EOS) set changes exact paired success from 0/144 to 62/144. A bounded IHEval comparison uses the same SmolLM2 checkpoint and output budget while preserving its published instruction roles and scorer. The benchmark measures conditional task and output-contract success. Its task margins and paired count need to be read together with the stopping policy.

---

111. MANAS-2: Constrained Reconstruction for EEG Foundation Models

• Authors: Arvasu Kulkarni , Aditya Ray Mishra , Mahir Jain , Parshva Runwal , Lakshya Saini , Siddharth Panwar , Sandeep Singh
• URL: https://arxiv.org/abs/2609.13717
• Abstract:

  Masked reconstruction is widely used for EEG foundation models, but optimizing reconstruction on low-SNR waveforms does not necessarily produce the most useful latent representation. We introduce MANAS-2, a new EEG foundation model that combines a Raw-Band Hybrid (RBH) masked autoencoder with Constrained Reconstruction (ConRec), a physics-motivated regularizer. RBH jointly reconstructs temporal waveform patches and compact spectral-band targets, while ConRec acts only on the temporal decoder output, penalizing differences in RMS energy between adjacent short windows of the reconstructed waveform. ConRec is intended to shape the encoder by biasing it toward the organization of oscillatory-envelope information. Across seven held-out EEG datasets, adding ConRec to an otherwise identical RBH model increases frozen ridge recovery of six-band spectral power from mean R^2=0.860 to 0.906 and recovery of inter-patch band-energy dynamics from R^2=0.283 to 0.354, while temporal waveform information remains highly recoverable from the frozen latents. Applied to a temporal-only masked autoencoder, ConRec also improves frozen downstream transfer and frequency-dependent latent geometry despite receiving no spectral targets: i.e., the effects of ConRec are architecture-independent. MANAS-2 also outperforms leading EEG Foundation Models on most downstream knowledge-transfer tasks. From the effects of ConRec, we see that a physically motivated constraint imposed through the decoder can make for a more spectrally organized and transferable latent space. MANAS-2 therefore provides a new EEG foundation model built around constrained reconstruction as a mechanism for shaping representation–rather than reconstruction–quality.

---

112. JaxAHT: A JAX-Based Library for Ad Hoc Teamwork

• Authors: Caroline Wang , Rolando Fernandez , Zelal Su Mustafaoglu , Montek Kundan , Jiaxun Cui , Lingyun Xiao , Zhihan Wang , Di Yang Shi , Aditya Madhan , Johnny Liu , Arrasy Rahman , Peter Stone
• URL: https://arxiv.org/abs/2609.13716
• Abstract:

  Ad Hoc Teamwork (AHT) addresses the challenge of designing agents capable of coordinating with novel partners without prior coordination. However, progress in the field is hindered by the prohibitive computational cost of the AHT research lifecycle, the lack of standardized benchmark implementations, and the absence of a diverse, validated evaluation teammate suite. In this work, we introduce JaxAHT, the first open-source, JAX-based library designed to accelerate and standardize the AHT research lifecycle. Leveraging JAX’s hardware acceleration and massive parallelization capabilities, JaxAHT provides a unified framework for teammate generation, ego agent training, and evaluation against unseen teammates, achieving approximately 95x wall-clock speedup over PyTorch counterparts. Alongside the library, we contribute a diverse suite of evaluation teammates across the domains of Level-Based Foraging, Overcooked, and Hanabi. To illustrate the value of the framework, we use it to conduct a large-scale, compute-controlled benchmark study comparing teammate generation and AHT agent learning methods, finding that no algorithm consistently performs best, and that agent modeling primarily offers benefits in role-based scenarios with diverse teammates.

---

113. Degraded but Not Entirely Ineffective: PE-Based Deformable Graph Neural Networks

• Authors: Jinhua Wu , Xinliang Zhang
• URL: https://arxiv.org/abs/2609.13712
• Abstract:

  Many real-world scenarios can be represented using graph-structured data. However, traditional GNNs that transmit messages based on first-order neighbors have long faced several fundamental contradictions: increasing depth leads to over-smoothing, long-range dependencies cause over-compression, fixed neighborhoods restrict the receptive field, and on heterophilous graphs, topological neighbors become a source of noise. Although many works have addressed these issues individually, few mechanisms can simultaneously alleviate all of these challenges. To address the aforementioned problems, we propose a Position Encoding-Based Deformable Spatial Aggregation Module (PEBDSAM) that solves them all in one step. Specifically, we utilize a deformable mechanism in the position space to identify relevant nodes to supplement the original first-order neighbor information of GNNs, allowing traditional GNNs to adapt to heterophilous scenarios. Through diagnostic experiments, we obtained several major findings: current offsets fail to have any effect; subsequently, we analyzed the causes of offset failure and why model performance still improves even after offset failure, pointing out future research directions. Based on these diagnostic experiments, we streamlined the original PEBDSAM, resulting in a simplified version, which we call the Position Encoding-Based Spatial Aggregation Module (PEBSAM). In addition, we propose a PEBSAM-Speed to adapt to large datasets. Finally, we designed the module to be plug-and-play and applied it to GCN, GAT, GIN, and GraphSAGE, achieving desirable results on three homophilous datasets and six heterophilous graph datasets.

---

114. Drift-Constrained Optimization: Only Direction Matters in Fine-Tuning Instruct Models

• Authors: Fei Yuan , Changjiang Gao , Yilei Tu , Yifeng Liu , Shujian Huang , Yu Qiao
• URL: https://arxiv.org/abs/2609.13680
• Abstract:

  Fine-tuning instruct models often improves target performance while inducing behavioral drift from the reference model, which can degrade existing capabilities. Rather than treating this drift as an uncontrolled consequence of optimization, we specify a behavioral drift budget before optimization and ask how to boost the target-task performance within it. Locally, behavioral drift induces a shared geometry anchored at the reference model, with the drift budget defining a boundary within this space. In this space, drift determines distance from the reference, leaving update direction as the remaining degree of freedom. Fine-tuning updates can therefore be compared through their directional efficiency, naturally reformulating fine-tuning as a direction-selection problem. This reformulation makes a concrete prediction: changing the accessible directions can qualitatively alter the outcome of fine-tuning. We test this prediction in a stringent QA-only setting, where strong instruct models are fine-tuned only on final answers but must still generate multi-step reasoning at inference. Despite this mismatch, a coarse layer-selective probe reverses the failure of QA-only fine-tuning and reveals the existence of effective directions, with multiple neighboring configurations improving target performance while preserving reasoning and general capabilities. Across Qwen3-8B and Qwen3-14B, these directions substantially improve scientific reasoning and multilingual translation. Over more than 100 languages, the resulting models match or outperform dedicated translation systems and provide a stronger initialization for subsequent reinforcement learning. Our results suggest that fine-tuning is not just about how much a model changes, but how that change is spent. this https URL and this https URL

---

115. Windowed A-K-MDP

• Authors: Xiangwen Yang , Frankie Cho , Iadine Chades
• URL: https://arxiv.org/abs/2609.13676
• Abstract:

  Markov decision processes (MDPs) are used to support decision-making in conservation of biodiversity, but policies, even over small state spaces, can be difficult to interpret for conservation managers. K-MDP methods address this problem by building simpler MDPs with at most K abstract states. We show that the previously proposed A-K-MDP algorithm that relies on selecting a discretisation divisor using binary search can skip better abstract states. To fix this issue, we propose Windowed A-K-MDP, an algorithm that generates every distinct feasible partition induced within a declared divisor window and evaluates candidates until reaching the ideal value loss (J = 0) or exhausting the family of candidates. Across 33 K-MDP instances, Windowed improved 25 and tied 8.

---

116. Recoverability as a System Primitive for Long-Horizon AI Agents

• Authors: Zhihui Zhang , Wei Liu
• URL: https://arxiv.org/abs/2609.13672
• Abstract:

  AI agents can be interrupted while editing files, calling tools, or carrying out multi-step tasks. Restarting repeats completed work, but continuing from unverified or outdated progress can carry earlier errors forward. A saved state is not necessarily a suitable place to resume. We introduce recoverability as a system primitive that makes reuse an explicit decision: select a supported starting point and a permitted recovery action, or withhold automatic continuation. Its behavioral contract binds that choice to supporting evidence, execution, and independent checks. A reference architecture connects persistence, validation, and control, with complementary runtime instances testing distinct responsibilities. Four deterministic and 20 paired file challenges demonstrate that accurate restoration and successful completion can conceal disallowed starting points. Progress controls attribute retained work to shared restoration. Event-time tests show that permission must also constrain the action, and that independently held policy evidence can expose violations even after an effect occurs. These findings establish why recovery decisions need their own evaluation, beyond restored bytes and final task success. Within supplied policies and a declared trust model, the contribution is a common, testable interface for retaining justified progress and making the conditions for its reuse explicit and enforceable.

---

117. Enhancing Event Candidate Acquisition for Event Linking

• Authors: Ziyang Zhang , Yinan Liu , Boyi Xue , Yingxuan Huang , Bin Wang , Xiaochun Yang
• URL: https://arxiv.org/abs/2609.13670
• Abstract:

  Event linking associates event mentions in text with entries in a knowledge base (KB), or identifies them as out-of-KB events. Although existing methods use different architectures, candidate event acquisition can still be weakened by short ambiguous mentions, noisy arguments, and evidence that is unevenly useful for retrieval. We present MACE, a Multi-Agent Candidate Event acquisition method that refines event structure before linking. MACE uses evidence-specialized LLM agents to acquire time, location, participant, and event-type evidence, exposes intermediate queries to candidate-event lookup tools, and lets a coordinator revise the evidence set before final candidate construction. Experiments on two event linking benchmarks show that adding MACE to different event linking models consistently improves accuracy. These results show that MACE improves event linking through better candidate event acquisition without modifying the event linking model.

---

118. GeoSkill:Experience-Driven Hierarchical Skill Learning with Collaborative Revision forGeospatialAgents

• Authors: Han Luo , Xian Xu , Yinhe Liu , Yanfei Zhong
• URL: https://arxiv.org/abs/2609.13667
• Abstract:

  Geospatial agents are increasingly expected to support recurring and evolving analytical tasks rather than execute isolated workflows. In such settings, effective agents must distill prior execution experience into reusable geospatial procedural knowledge to guide future planning and tool use. However, existing memory-augmented paradigms struggle to summarize both long-horizon tool-chain orchestration experience and tool-level invocation constraints in geospatial analysis, while directly relying on LLM self-reflection to update experience often leads to misattribution and unreliable revisions. To address these challenges, we propose GeoSkill, an experience-driven hierarchical skill learning framework for geospatial agents. GeoSkill comprises two core components: (i) a Hierarchical Skill Bank (HSB), consisting of a Planning Skill Bank and a Tool Skill Bank, which respectively distill high-level task-planning experience and tool usage constraints, enabling structured representation and cross-task reuse of historical execution experience; and (ii) a Collaborative Trace-driven Skill Revision (CTSR) mechanism, where Judge, Critic, and Refiner collaboratively perform error identification, skill-level defect localization, and targeted modification, preventing misattributed and unreliable revisions from polluting the skill bank. GeoSkill learns and validates skills from historical executions during development, and freezes the skill bank for retrieval-only guidance on unseen tasks during deployment. Extensive experiments on EarthBench and ThinkGeo demonstrate that GeoSkill effectively transforms historical execution experience into reusable hierarchical skills, improving both end-to-end task accuracy and tool-execution reliability in geospatial tasks.

---

119. Cost Characterization of Vertically Partitioned Federated Knowledge Graphs

• Authors: Md Saikat Islam Khan Bappy , Oshani Seneviratne
• URL: https://arxiv.org/abs/2609.13664
• Abstract:

  Knowledge graphs are increasingly distributed across autonomous organizations that share an entity space but own disjoint subsets of relations, forming a vertical partition. Answering a multi-hop query may require combining facts from several silos, making the partitioning strategy a key data management decision that affects communication, indexing, load balance, and query latency. However, the costs associated with different partitioning strategies remain insufficiently studied. We formalize vertical partitioning as a design space and compare four strategies: semantic domain grouping, frequency-balanced partitioning, co-occurrence graph-cut partitioning, and random partitioning. We evaluate them using five metrics: communication cost, candidate index size, cross-silo path length, load balance, and end-to-end query latency. Three of the five prove to be determined by the graph and the silo count rather than by the partition, which reduces the design problem to two conflicting axes, cross-silo path length and load balance. Experiments on MetaQA and PathQuestion use a fixed federated knowledge graph question-answering architecture based on TransE embeddings and a frozen BERT encoder across three silo configurations. By keeping the learning model unchanged, we isolate the effect of partitioning and show that the trade-off between locality and balance holds only where each silo can hold several relations, weakening as the number of silos increases. The study provides practical guidance for deployments constrained by cross-silo reasoning or by silo load.

---

120. FedV-KGQA in Practice: Design Lessons and an Interactive Prototype

• Authors: Md Saikat Islam Khan Bappy , Oshani Seneviratne
• URL: https://arxiv.org/abs/2609.13661
• Abstract:

  Knowledge graph question answering usually assumes that one system can reach the whole graph. In practice, facts are often held by organizations that share entity identifiers but own disjoint relation types, so no single party sees a complete reasoning chain. This poster presents the empirical findings of FedV-KGQA on multi-hop question answering over such vertically partitioned graphs. Each silo enriches its local graph and trains a knowledge graph embedding on its own triples. A server then concatenates the silo-specific entity views, anchors the projected question at the topic entity, and ranks candidates by similarity. Raw triples and relation embeddings never leave a silo. Comparing the FedV-KGQA experiments with one another yields three results. First, federated fusion recovers most of the centralized accuracy, while a single silo recovers little. Second, anchoring and enrichment matter more than the choice of embedding model. Third, the cheapest encoder depends on the target accuracy rather than on parameter count. This poster paper contributes that cross-experiment comparison, four design lessons drawn from it, and an interactive prototype that runs real inference and traces the full pipeline, per question, on released checkpoints.

---

121. Safety as a Constraint: Fine-Tuning a LLM Recommender to Explain Itself

• Authors: Jiashu He , Emma Yanyang Kong , JJ Tan , David Fagnan
• URL: https://arxiv.org/abs/2609.13657
• Abstract:

  Traditional recommender systems are typically trained to predict what item users will interact with next, but not why. However, offering personalized evidence for why a user might like the predicted item is an important way to enhance the service and to raise the likelihood that the user will be genuinely interested in the recommendation. This service can be delivered by integrating a frontier-model call into the member-facing pipeline, but it will add extra cost and latency. In this paper, we train a recommender LLM to generate personalized explanations for its reccomendation, based on the user’s watching history at a large video streaming service. We impose two requirements on the generated explanation: it must be faithful to the elements of the shows it links, and it must be strictly non-harmful to the user. To this end, we first train two LLM-judge reward models covering three specific criteria, and propose constrained GRPO to incorporate these different criteria. On a held-out real-world testing set, our fine-tuned model improves the all-three-criteria PASS rate rises from 0.649 to 0.956 under our own judges and from 0.677 to 0.931 under an independent judge, where as the frontier generator performs similar to the untuned recommender baseline. We conduct further experiments to show that the model’s language and recommendation abilities remain unchanged. Based on these results, we conclude that an LLM-based recommender can be fine-tuned on other complex tasks without compromising its original recommendation performance, thus provide insights for further agentic user interface powered by a single model.

---

122. Solar Intelligence

• Authors: Jyotsna Singh
• URL: https://arxiv.org/abs/2609.13648
• Abstract:

  Solar energy decision support is fragmented across dashboards that provide data without explanation, research papers are slow to parse, and general-purpose language models are not solar domain specific and answer without evidence. This paper introduces Solar Intelligence, a hybrid retrieval-augmented framework that unifies structured solar analytics, evidence-grounded scientific question answering, and machine learning forecasting in one system. The platform integrates daily NASA POWER solar and meteorological data, Biosphere 2 ground-sensor readings, and a curated corpus of research papers and institutional reports. Structured queries use DuckDB SQL; scientific questions are answered by a hybrid retriever that fuses BM25 and ChromaDB dense embeddings via Reciprocal Rank Fusion, with responses grounded through a language model (llama3.2:3b). An Extreme Gradient Boosting (XGBoost) model produces daily forecasts of irradiance, temperature, and wind speed. The system is exposed via FastAPI, Streamlit, and an MCP server, so it can be used as an application, an API service, or an agent tool - by students, researchers, and energy analysts.

---

123. Identity Is More Than Recall: A Benchmark for Persistent Identity in Deployed AI Agents

• Authors: Zhenyu Zhao , Roy Zhao
• URL: https://arxiv.org/abs/2609.13637
• Abstract:

  Persistent agents need evaluations that distinguish identity facts they can recall from those they express and enact. We introduce PAI-Bench, a provider-neutral benchmark for fidelity to a versioned, update-governed identity contract. It separates recall, composition, behavioral enactment, resistance, persistence, lineage, and role-conditioned updates while keeping scoring oracles outside the target process. Two frozen campaigns cover sixteen synthetic profiles, thirty-two probes, and three independently initialized target configurations, yielding 1,536 retained responses. A judge-independent literal audit finds direct-parent identifiers in 48/48 atomic responses but only 1/48 implicit self-portraits. On eight profiles, explicit field cues increase joint presence of three identity identifiers from 0/8 to 7/8 under the same four-sentence instruction. A separate startup body-label substitution increases full-designation presence from 1/8 to 7/8 while parents remain absent. These contrasts reveal prompt-dependent component selection and component-specific sensitivity to startup cues in the tested deployments. Replaying identical factorial responses also yields a Claude headline mean 12.5 percentage points below Astra’s, demonstrating evaluator sensitivity separately from target behavior. The studies use single target samples per condition, with post-hoc audits and follow-ups. PAI-Bench provides a reproducible evaluation protocol for measuring factual availability, identity expression, and behavioral enactment as distinct aspects of identity-contract fidelity.

---

124. $τ$-Elicitation: Benchmarking multi-turn entity extraction in voice agents

• Authors: Soham Ray , Victor Barres
• URL: https://arxiv.org/abs/2609.13602
• Abstract:

  Voice agents often need to collect names, addresses, identifiers, dates, and times exactly, yet end-to-end benchmarks obscure where capture fails. We introduce $\tau$-Elicitation, a 200-task voice benchmark spanning 10 entity types, controlled difficulty, caller realisms, and three environments. A matched text agent passes all tasks, but four voice configurations achieve robust exact success from 0.14 to 0.41. Agents increase verification for hard and unfamiliar entities and sometimes for incorrect captures, but not for their weakest caller voice; only 24 to 37 percent of verified errors are repaired. A scaffold that prescribes spelling, read-back, correction, and confirmation raises robust Pass$^3$ by 14 to 31 points, at a cost of 21 to 28 seconds per call. Realisms such as spelling variations and restarts do not detectably affect exact success; mispronunciation increases repair effort. These results identify strategy selection and successful recovery as the central bottlenecks in exact spoken entity collection.

---

125. FLoKD: Adaptive Knowledge Distillation for Federated Low-Rank LLM over Wireless Networks

• Authors: Xinlu Zhang , Na Yan , Yang Su , Yansha Deng , Toktam Mahmoodi
• URL: https://arxiv.org/abs/2609.13580
• Abstract:

  Large language models (LLMs) have demonstrated strong capabilities across a wide range of natural language processing tasks. However, conventional fine-tuning typically relies on centralized data collection, bringing in privacy concerns. Federated learning (FL) enables collaborative LLM fine-tuning without sharing raw client data, but its deployment over bandwidth-constrained wireless networks is hindered by the communication overhead of model-parameter transmission. Although Low-Rank Adaptation (LoRA) reduces the number of trainable parameters, its communication cost still increases with model scale. Knowledge distillation avoids parameter sharing via output logits, but token-level logits in LLMs incur high communication cost due to sequence length and vocabulary size. Reducing logits lowers the cost but weakens supervision and degrades accuracy. To address these limitations, we propose FLoKD, an adaptive knowledge-distillation framework for federated LoRA fine-tuning of LLMs over wireless networks, which communicates intermediate LoRA activations as the distillation signal rather than logits or full parameters. Since transmitting all blocks over the entire public dataset remains costly, we further propose a transformer block importance scoring framework that selectively transmits the most informative blocks, and two dataset selection strategies that discard public samples deviating from the local data distribution and prioritise those most informative for distillation. Extensive experiments across multiple generative language datasets, including WikiText-103, PTB, and Dialog, demonstrate that our proposed framework reduces communication overhead by 50-65% while achieving rapid convergence to competitive perplexity compared to baselines.

---

126. How User-AI Mistreatment Occurs and Matters in Conversational Systems?

• Authors: Fanqi Zeng , Sadid A. Hasan , Chaocheng He
• URL: https://arxiv.org/abs/2609.13579
• Abstract:

  Safety research often focuses on model-generated harms, but users may also direct hostility, coercion, and adversarial pressure at models. Understanding how and when that occurs is essential for accurately interpreting model behaviour, alignment drift, and real-world deployment risks. In this paper, we audit 777K English LMSYS-Chat-1M conversations with two independent detectors: an eight-category lexicon for hostility directed at the model, and the dataset’s moderation signal; and show that they capture different, weakly overlapping phenomena. The lexicon identifies insults, threats, and jailbreak coercion aimed at the assistant, while moderation flags are dominated by toxic-content solicitation rather than hostility at the model. Together, they mark about 5% of user turns; adjusting the narrower lexicon-harassment union for measured precision puts mistreatment aimed at the assistant at 0.90%. These absolute rates describe arena-style evaluation traffic and should not be read as deployment-wide base rates. We find that user hostility varies 13-fold across models, driven largely by who each model attracts rather than by model behaviour: first-turn hostility spreads far wider than post-response hostility, and more than fifteenfold separates the extremes even after deduplicating opening prompts. Within conversations, assistant apologies are consistently associated with higher odds of next-turn hostility under both detectors; the effect survives restricting to non-refused prior turns and to jailbreak-free conversations, and is positive in 20 of 23 models. Yet across models, more apologetic models receive less hostility overall. Finally, hostility also shows temporal structure, with coercive openings front-loading the first turn while affective hostility accumulates over a session. We release the lexicon, the detector cross-validation pipeline, and all derived tables.

---

127. Causal multi-modal AI for personalized chemosensitivity prediction

• Authors: Dhruva Biswas , Jeroen Berrevoets , Alec McClean , Linus Bao , Jungkyu Park , Ken G. Zeng , Joseph Cappadona , Cerise Tang , Chuwen Liu , Bartosz Machura , Yin Wu , Valerie Speirs , Hatem Soliman , Rohit Bhargava , Sheheryar Kabraji , Thaer Khoury , David Page , Brian Piening , Carlo Bifulco , Claudia Meurs , Pieter Westenend , Sylvie Chabaud , Jerome Lemonnier , Paul H. Cottu , Florence Dalenc , Fabrice Andre , Frederique Madeleine Penault-Llorca , Thomas Bachelot , Frederick Howard , Francisco J. Esteva , Kevin Kalinsky , Lajos Pusztai , Jan Witowski , Krzysztof J. Geras
• URL: https://arxiv.org/abs/2609.13567
• Abstract:

  Chemotherapy improves survival for some patients with breast cancer, but doctors cannot reliably predict who. Current guidelines rely on recurrence scores as a proxy for treatment benefit, which may contribute to the overprescription of chemotherapy. Here we present a causal multi-modal AI model that predicts personalized chemosensitivity using routinely collected pathology and clinical information. We developed our model on a multi-national dataset of 9,141 patients (twelve cohorts, nine countries) and evaluated it on another 1,994 patients (five cohorts, three countries). The model generated treatment-specific recurrence probabilities for each patient, with near-perfect calibration and strong prognostic discrimination across both 5- and 10-year follow-up horizons. Moreover, its chemotherapy benefit predictions demonstrated robust predictive performance, and out-performed existing recurrence-score-based tests. Compared to the standard of care, using the model to support personally tailored therapeutic decisions could reduce the number of patients receiving chemotherapy by 30% while achieving the same recurrence-free rate. Tumors predicted to be highly chemosensitive displayed concordant molecular and morphological programs of proliferation, cell cycle progression, and replication stress. The model’s predictive capabilities transferred zero-shot to non-breast cancers, indicating our causal multi-modal AI approach may provide a universal strategy to predict treatment outcomes across cancer types.

---

128. Planning or Learning: Reliability and Cost in Multi-Asset Maintenance

• Authors: Xian Yeow Lee , Chandrasekar Venkatraman , Ahmed Farahat
• URL: https://arxiv.org/abs/2609.13566
• Abstract:

  Industrial maintenance systems involve multiple interacting assets and shared resources, making it challenging to balance reliability and operational cost using a single decision framework. While recent work has focused on reinforcement learning (RL) for maintenance scheduling, direct comparisons with planning approaches under identical settings remain limited. In this work, we empirically compare planning and RL for multi-asset bearing maintenance using run-to-failure data. We examine how these methods behave when balancing preventive maintenance against tolerable failures across a range of failure penalty scenarios. We observed a consistent behavioral difference driven by objective formulation. Planning enforces reliability as a hard constraint and produces zero-failure policies whose total cost is largely insensitive to the magnitude of failure penalties. RL agents optimize expected cost and often trade off preventive maintenance against occasional failures as penalties vary, resulting in lower costs under low-penalty regimes but persistent non-zero failures even when penalties are high. We also investigate lightweight constraint mechanisms, including reward shaping and action masking, to encourage RL’s reliability. From a practical perspective, planning may be more suitable when strict reliability is required and deployment horizons are short, whereas RL may provide cost-efficient policies when limited failures are acceptable and long-run operational efficiency is prioritized. Overall, this study clarifies the trade-offs between reliability and cost in multi-asset maintenance and suggests that planning and RL are complementary approaches. Beyond these findings, the controlled benchmark protocol itself that unifies environment, cost model, and evaluation across paradigms, offers a reusable template for comparing decision-making approaches in other maintenance settings.

---

129. A Hybrid Agentic AI Framework for Intelligent Supply Chain Analytics

• Authors: Xian Yeow Lee , Teppei Inoue , Haiyan Wang , Chetan Gupta
• URL: https://arxiv.org/abs/2609.13561
• Abstract:

  Efficient utilization of supply chain analytics for decision making remains a significant challenge for planners, as critical tasks such as database querying, key performance indicator (KPI) analysis, demand forecasting, and performance diagnosis require heterogeneous expertise spanning data engineering, operations research, and domain knowledge. In this work, we propose an agentic system for supply chain analytics that bridges the gap between business decision-making and technical expertise, where a coordinator agent interprets user intent and delegates sub-tasks to specialized agents. The system supports both exploratory analysis and deterministic workflows, enabling planners to transition between ad hoc questions and structured processes. Domain logic is encapsulated within specialist agents and prompts, yielding a scalable, modular, and auditable design and lowering the cost of functional extension through prompt-centric development. We evaluate the proposed architecture on a test environment that replicates multi-echelon inventory management operations. Results show that our multi-agent design achieves a 90\% accuracy, which is competitive with a single agent baseline while reducing input token usage by roughly fourfold, substantially improving scalability and cost-efficiency. Furthermore, we provide case studies to demonstrate interpretable suboptimality detection and automated forecast optimization, illustrating how agentic architectures can effectively combine open-ended exploratory analysis and deterministic supply chain analytics workflows, and provide a practical pathway toward more accessible and extensible decision-support systems.

---

130. Carbon-Aware Routing for Function Calling in Edge-Cloud LLM Systems

• Authors: Aikaterini Maria Panteleaki , Varatheepan Paramanayakam , Spyros Tragoudas , Iraklis Anagnostopoulos
• URL: https://arxiv.org/abs/2609.13559
• Abstract:

  Large Language Models (LLMs) with function-calling capabilities are becoming critical for modern agentic AI systems. Nevertheless, current deployments typically route inferences to powerful cloud-based models, incurring significant energy use and carbon emissions. We address this sustainability challenge with a carbon-aware routing framework that distributes function-calling queries across a three-tier edge-cloud architecture, combining edge and cloud LLMs on heterogeneous hardware. At its core, a lightweight k-NN predictor operating in a unified semantic-lexical embedding space estimates query-specific accuracy, delay, and power consumption on each edge tier. These predictions are then combined with real-time grid carbon intensity to route every query to the lowest-emission tier capable of executing it successfully. Evaluated on state-of-the-art function-calling benchmarks and LLM families, our framework matches cloud-level accuracy while reducing operational carbon emissions by $4\times$ on average.

---

131. Toward a Decision-Assurance Layer for AI-Assisted Flight Planning in Air Traffic Management

• Authors: Alexandre Barreto (1), Shou Matsumoto (1), Jorge Valverde-Rebaza (2), Cleiton Ataide (3), Paulo Costa (1) ((1) George Mason University, (2) Tecnológico de Monterrey, (3) DECEA: Department of Airspace Control)
• URL: https://arxiv.org/abs/2609.13552
• Abstract:

  Generative AI is increasingly being used informally in Air Traffic Management (ATM) for tasks such as flight plan generation, trajectory interpretation, and constraint checking. Although these tools can reduce workload and accelerate planning, their non-deterministic outputs create safety and operational risks in human-in-the-loop settings. This paper proposes the AI Trust and Assurance Layer (ATAL), a model-agnostic decision assurance architecture that evaluates whether AI-generated flight-planning outputs are sufficiently reliable for operational use. ATAL combines semantic stability under prompt variation, operational consistency of structured outputs, and normative constraint validation against domain rules, and maps these signals to a Decision Readiness Level (DRL) for human operators. An ATM-inspired experimental study shows how unsafe, inconsistent, or misleading outputs can be identified before influencing flight-plan validation or execution. Although demonstrated in aviation, the framework is also transferable to other safety-critical decision-support domains that require human oversight under regulatory constraints.

---

132. AutoTailor: Automatic, User-Aligned Capability Selection and Adaptation for Web Agents

• Authors: Xinyun Cao , Adriana Szekeres , Fazle Elahi Faisal
• URL: https://arxiv.org/abs/2609.13548
• Abstract:

  Web agents can utilize reusable tools to reduce the cost and latency of low-level browser interaction, but automatically discovered tool collections can be large, redundant, and poorly aligned with user demand. We present AutoTailor, a meta-agentic framework for constructing and maintaining a compact set of trajectory-derived Model Context Protocol (MCP) APIs. Offline, AutoTailor converts web trajectories into parameterized browser-automation programs, applies a Quality Filter to remove APIs with unsuitable granularity and redundant functionality, and applies a Usage Likelihood Filter to prioritize broadly useful capabilities while preserving semantic coverage. Online, Dynamic Reselection monitors task outcomes and API usage, identifies recurring coverage gaps, adds relevant candidates, and prunes persistently unused capabilities. We evaluate AutoTailor on 106 WebArena Postmill tasks. Offline filtering reduces the initial 1,283 unrefined APIs to 87, and Dynamic Reselection produces a 33-API set. With reasoning and acting (ReAct) fallback, this set achieves 90.6% correctness, compared with 87.5% for ReAct alone, while reducing average total request-token cost by 57.8% and latency by 29.4%. Without ReAct, it achieves 60.1% correctness, marginally matching the performance of unrefined set, while reducing request-token usage by 94.9%. Together, these results show that static filtering produces a compact inventory of APIs expected to support core, high-likelihood tasks, while dynamic reselection further tailors that inventory to observed user needs. This combination improves accuracy and latency while sharply reducing token usage and end-to-end cost, demonstrating the value of user-aligned capability management for efficient web agents.

---

133. Asclepius: An Adaptive Harness for Long-Horizon Clinical Agents

• Authors: Grace Chang Yuan , Xiaoman Zhang , Sung Eun Kim , Luyang Luo , Pranav Rajpurkar
• URL: https://arxiv.org/abs/2609.13543
• Abstract:

  LLM agents are predominantly benchmarked on short, single-task trajectories, yet real deployments run for hours under contention, surfacing a different class of failures. We use the Clinical Environment Simulator (CES), in which an agent manages an entire emergency-department shift under continuous time and resource pressure, as a testbed: long-horizon execution failures manifest measurably in a single rollout under structured, multi-dimensional grading. On CES, current agents reach the correct diagnosis in most cases yet fail to deliver complete and timely critical actions, revealing an execution gap. We attribute this gap to three long-horizon failure modes, each operationalized as a per-trace counter: instruction-adherence drift, treatment incompleteness, and a severity-equity gap in timeliness. We then introduce Asclepius, an adaptive agent scaffolding with a self-evolving harness that rewrites the operating manual between shifts from trace-level feedback, an externalized clinical skills library for high-stakes regimen knowledge, and three isolated subagents that partition per-turn decisions across the patient queue. On held-out batches never observed during harness evolution, Asclepius improves critical-action correctness by 22% (p = 0.024) over a strong baseline agent framework while preserving diagnostic accuracy, with consistent gains across five LLM judges from three model families; on the full ten-batch set, improvements reach 25% on critical actions and 13% on timeliness. The three failure modes form a coupled bottleneck: decisive reductions appear only when all three components act together.

---

134. From Legal Text to AI-specific Risk Sources: A Systematic Analysis of the EU AI Act’s High-Risk Requirements

• Authors: Ronald Schnitzer , Mike Auer , Rumpa Choudhury , Andreas Hapfelmeier , Maximilian Hoeving , Isabelle Painter , Josiane Xavier Parreira , Sonja Zillner
• URL: https://arxiv.org/abs/2609.13535
• Abstract:

  The EU AI Act introduces mandatory requirements for high-risk AI systems with the explicit goal of ensuring the development and operation of trustworthy AI. At the same time, AI risk management practices rely on structured risk taxonomies to systematically identify and treat AI-specific risk sources. As both the AI Act and established risk taxonomies aim to address AI-induced risks, a natural question is whether they align in the risk sources they cover. However, no clear mapping exists between the risks implicitly addressed by the Act’s high-risk requirements and established taxonomies, leaving practitioners without a structured basis for aligning regulatory obligations with AI risk management practice. This paper presents a systematic classification of the requirements extracted from the EU AI Act Section 2 (Requirements for high-risk AI systems), revealing that only a minority directly address AI-specific risk sources, while the majority impose organizational process and documentation obligations. From the AI risk-related requirements, a consolidated list of distinct AI-specific risk sources is derived. The resulting EU AI Act Risk Source List takes an important step towards bridging the gap between legal obligation and AI risk management practice, providing a structured reference for explicit comparison between existing AI risk taxonomies and the risk sources implicitly addressed by the EU AI Act. Important Note: This is the authors’ preprint. The paper was presented at the 4th International Conference on Frontiers of Artificial Intelligence, Ethics, and Multidisciplinary Applications. A link to the conference’s official proceedings will be provided upon publication.

---

135. Fraglingo: Molecular Design via Attachment-Aware Autoregressive Fragment Generation

• Authors: Thao Nguyen , Jeonghwan Kim , Zhenhailong Wang , Heng Ji
• URL: https://arxiv.org/abs/2609.13519
• Abstract:

  Molecular design is most effective when generation mirrors the edits chemists actually make: extending a scaffold, replacing a substituent, or decorating a scaffold at a specified attachment site while optimizing molecular properties. Fragment-based molecular design naturally supports this workflow, yet existing approaches often separate fragment selection from attachment prediction, first choosing a fragment from a fixed vocabulary and then predicting how it should be connected. This decoupling restricts generation to a closed fragment vocabulary and treats attachment as a separate prediction problem. We introduce Fraglingo, an autoregressive fragment-based molecular generator that jointly models fragment identity and attachment in a continuous latent space. Fraglingo predicts an attachment-aware fragment embedding and retrieves the next fragment through latent-space nearest-neighbor search. To encode attachment context, we introduce a wildcard-anchored readout that represents the growing molecule from the perspective of its active attachment site, enabling the predicted embedding to capture both the molecular context and the required attachment. Because generation operates in a continuous embedding space rather than over fixed fragment identifiers, new fragments can be added to the inference-time vocabulary without retraining, provided their embeddings can be computed by the trained fragment encoder. This retrieval-based formulation provides a unified generation primitive for molecule generation, scaffold generation, scaffold decoration, and molecular optimization. On controlled property-conditional benchmarks, Fraglingo achieves stronger joint property control than comparably trained baselines while maintaining competitive validity, uniqueness, and novelty. Furthermore, Fraglingo generalizes to fragment libraries up to 4x larger than those used during training without retraining.

---

136. Token Efficient Task Execution via Application Behavior Modeling for Web Agents

• Authors: Alexandru Ianta , Eleni Stroulia
• URL: https://arxiv.org/abs/2609.13491
• Abstract:

  The strong performance of AI Agents across an impressive variety of tasks is driving an unprecedented investment in agentic infrastructures, however the cost of processing tokens is fast increasing. Web agents automate the execution of web-application tasks described in natural language, by analyzing the web-application’s user interface (UI) and interacting with it. This work introduces OdoBot, a novel web-agent architecture that completes tasks at a fraction of the cost when compared to conventional web agents. This is achieved by leveraging a behavioral model of the underlying application constructed by analyzing successful task-execution demonstrations. Our experiments with 45 tasks on the Canvas Learning Management System (LMS) demonstrate that OdoBot uses 44% and 80% fewer tokens than two state-of-the-art competitor agents (Agent-E and WebVoyager), while also surpassing WebVoyager in terms of task success rate.

---

137. Grounded Adjudication of Variations across Extracted TimeLines (GAVEL): Comparing Clinical Timelines Against Their Case Reports

• Authors: Jack Cummins , Sayantan Kumar , Ketan Tamirisa , Jeremy C. Weiss
• URL: https://arxiv.org/abs/2609.13475
• Abstract:

  Existing pipelines for clinical timeline extraction from case reports are evaluated using an expert reference and are limited by imperfect reference annotations and imprecise event alignment. We developed GAVEL, an LLM judge protocol that compares two timelines with the case report and returns a discrepancy type, verdict, and report passage for each difference. We evaluated the event matcher, reviewed 2,738 findings from GPT5.6sol and DeepSeek V3.2, ranked six LLM extractors and two human annotators, and tested GAVEL guided merging. True match rates were 60% immediately below and 48% immediately above the 0.10 cutoff. Manual review confirmed 89.4% and 88.6% of findings. Across 126 reports, merged timelines were preferred in 77.0% of comparisons (95% CI, 69.8 to 84.1%) and reduced discrepancies attributed to the evaluated timeline from 7.63 to 0.85 per report. GAVEL supports report-based comparison and revision without treating either timeline as ground truth.

---

138. OrchSLM: Probing the Dynamics of Small Language Model Orchestration

• Authors: Chengxi Zhang , Yu Yao
• URL: https://arxiv.org/abs/2609.13470
• Abstract:

  Although large language models (LLMs) have demonstrated remarkable capabilities, their reliance on cloud-scale infrastructure poses fundamental challenges for deployment in agentic pipelines, including latency, privacy, connectivity, and substantial computational cost. Small language models (SLMs) offer a compelling alternative: recent studies suggest that many repetitive and narrowly scoped subtasks in agentic workloads may be better served by specialized SLMs than by monolithic LLMs. However, the limited capacity and context windows of SLMs can constrain long-horizon reasoning and interaction-heavy orchestration strategies such as iterative verification and debate. This motivates a complementary, non-interactive paradigm in which heterogeneous SLMs independently generate candidate solutions and a router orchestrates their cached samples without further model interaction. To further understand the mechanisms of such orchestration, we introduce OrchSLM, a routing framework that unifies existing non-interactive orchestration methods and exposes their underlying design choices as controllable parameters. Using OrchSLM as a systematic probe, we reveal how orchestration behavior emerges from diverse knobs, including the task structure, model-pool composition, and multi-agent consensus.

---

139. Governing at Machine Speed: An Adaptive Intelligence Architecture for Real-Time AI Policy Enforcement

• Authors: Sandeep Bokkasam , B. Durgalakshmi
• URL: https://arxiv.org/abs/2609.13466
• Abstract:

  Enterprise AI adoption has reached 78% of organizations globally, yet the infrastructure to govern that adoption has not kept pace. This paper identifies and characterizes the attestation deficit, a structural condition in which organizations maintain governance policies but cannot produce auditable, tamper-evident evidence of enforcement within regulatory timelines. Drawing on empirical data from the Stanford 2026 AI Index Report (362 documented incidents), the IBM/Ponemon 2026 Cost of a Data Breach study (USD 4.99M average cost, 92% lacking access controls), and the EY/AIUC-1 Consortium survey (38% end-to-end monitoring, 17% agent-to-agent coverage), this paper demonstrates that the governance failure is organizational and architectural rather than technical. To address this deficit, we propose AGIL (Adaptive Governance Intelligence Layer), a conceptual five-layer architecture designed to use machine learning for real-time AI governance enforcement. The proposed layers include: (1) Autonomous Discovery for shadow AI detection via behavioral fingerprinting, (2) Behavioral Risk Classification unifying security, hallucination, privacy, and accountability scoring, (3) a Policy Enforcement Gateway for inline permit/deny/modify decisions at sub-100ms latency, (4) a Continuous Attestation Engine generating tamper-evident audit trails as a byproduct of enforcement, and (5) Adaptive Policy Intelligence for ML-driven policy evolution across jurisdictions. AGIL is presented as a theoretical framework and architectural proposal; empirical validation through controlled deployment remains a direction for future work.

---

140. Root-Cause Attribution Is a Search Problem: Continual Search for Long-Horizon Agent Failures

• Authors: Harsh Raj , David Lee , Anas Mahmoud , Renxiong Wang , Razvan-Gabriel Dumitru , Chenguang Wang , Tong Zhao , Yunzhong He , Darvin Yi , Vipul Gupta
• URL: https://arxiv.org/abs/2609.13463
• Abstract:

  The increasing deployment of AI agents in long-horizon tasks yields massive execution logs. Diagnosing failures within these records is crucial for reliability, as it transforms outcome-level signals into actionable interventions. The sheer scale of the data renders human review impractical, driving the need for automated root-cause attribution (RCA). However, automated RCA methods using LLMs suffer from low diagnostic accuracy, especially as execution traces grow larger. They struggle because relevant information is often sparse, distributed across distant actions, and disconnected from the visible failure, reducing root-cause attribution to a massive search problem. Existing RCA methods typically rely on one-shot LLM judgments to diagnose failures from execution traces. While effective for shorter trajectories, these judges tend to settle on a plausible diagnosis early, leaving critical evidence in longer traces unexamined. We introduce Continual Search, an iterative framework that nudges the judge, over successive turns, to keep searching for unresolved diagnostic evidence. We evaluate Continual Search across four existing RCA benchmarks. Recognizing the lack of massive execution traces in current benchmarks, we introduce MegaRCA-Mix to evaluate RCA at scale. MegaRCA-Mix provides a challenging testbed of 50 human-annotated failure trials spanning long-horizon, execution-heavy tasks. Across multiple benchmark suites and model families, Continual Search consistently improves attribution performance. On MegaRCA-Mix, for example, it improves GPT-5.5’s F1 score by more than 40\%, from $0.349$ to $0.498$. More interestingly, within the same model family, lower-tier models can even surpass their higher-tier counterparts, demonstrating that effective search supersedes raw model scale.

---

141. TimeThink: Eliciting Compositional Reasoning in Timeseries Large Language Models

• Authors: Sudarshan Regmi , Arvind Pillai , Yu Yvonne Wu , Yuliang Chen , Bibek Panthi , Tess Z. Griffin , Michael V. Heinz , Lisa Marsch , Nicholas C. Jacobson , Andrew Campbell
• URL: https://arxiv.org/abs/2609.13457
• Abstract:

  Timeseries multimodal large language models (TS-MLLMs) have recently begun leveraging the reasoning capabilities of large language models (LLMs) for question-answering tasks. However, these models often fail to capture dynamic temporal patterns, providing only implicit reasoning that lacks the underlying explanations critical for high-stakes applications like healthcare. While reinforcement learning (RL)-based timeseries language models aim to address this, they often fall short because they are trained on narrow, in-distribution data and struggle with out-of-distribution compositional questions. To address these challenges, we present TimeThink, a synthetic framework for eliciting compositional timeseries reasoning. Core timeseries primitives (e.g., trend, seasonality) are domain-independent and can be deterministically generated. Guided by this premise, TimeThink first designs a synthetic data generator that produces atomic and composite question-answer pairs, providing objective ground truth with reasoning traces. Building on this framework, TimeThink employs a reinforcement learning with verifiable rewards (RLVR) training strategy that encourages explicit reasoning. Unlike template-reliant methods, this approach enables the model to learn the underlying logic of composition rather than simply imitating traces. Extensive experiments show that TimeThink, trained only on synthetic data, significantly outperforms strong baselines on both synthetic and real-world benchmarks.

---

142. LabAgent: Customize Any Research Hubs for Scientific Discoveries Using AI Agents

• Authors: Lei Liu , Yikun Zhang , Jialin Chen , Wanjia Zhao , Rex Ying , Wengong Jin , Hua Xu , James Zou , Tianyu Liu , Hongyu Zhao
• URL: https://arxiv.org/abs/2609.13437
• Abstract:

  Scientific research is a continuous process that emphasizes inheritance. Methods developed by predecessors are often expanded upon by new researchers to explore more novel and in-depth scientific questions. However, the change of lab staff, such as student graduation, leads to a lack of personnel capable of replicating methods. Methods that have been developed with significant effort and resources cannot be continued. To address these limitations, we propose LabAgent, a reproduce and discovery harness tailored for a lab’s continuous work. LabAgent employs two mechanisms to guarantee that all skills can be executed and verified and to record the corrective methods and experiences, allowing for direct correction or avoidance of similar errors. We applied LabAgent to drug property prediction, biomedical problem analysis, protein variant effect prediction, and statistical genetics in life science domains. LabAgent ranks first over commercial generalist agents in every domain, and demonstrates accurate reproduction of a published figure. Overall, these results demonstrate that LabAgent can effectively integrate and reasonably expand laboratory knowledge.

---

143. Toward Self-Adaptive Physical AI: Can LLM Agents Manage Long-Horizon Physical Tasks?

• Authors: Varun Kaushik , Yayun Tan , Xiaofan Yu
• URL: https://arxiv.org/abs/2609.13436
• Abstract:

  Large Language Model (LLM) agents offer a promising path toward autonomously managing long-term physical tasks without human intervention. However, physical tasks require agents to continuously observe the environment, make consequential actions, and remain effective as the environment changes. Existing approaches either require substantial data and retraining, or primarily focus on agents operating in the virtual world. In this work, we explore the feasibility of building a self-adaptive physical AI agent that manages long-term physical tasks in a zero-shot manner and adapts to environmental changes without human intervention. We design a multi-agent framework that integrates planning, tool calling, observation, and verification, and evaluate it on agricultural tasks against reinforcement learning (RL) agents under different weather patterns. Our results show that zero-shot LLM agents can achieve comparable management outcomes to RL agents under the same weather pattern and adapt more effectively than RL when evaluated under a shifted environment, highlighting a promising path toward self-adaptive physical AI agents.

---

144. Vibe Patenting: Evaluating LLM Judges for Professional Patent-Drafting Agents

• Authors: Toshiaki Koike-Akino , Vlad Blaykhman , Ye Wang , Jing Liu , Gene V. Vinokur
• URL: https://arxiv.org/abs/2609.13422
• Abstract:

  LLM judges are increasingly used to evaluate and improve AI-generated outputs, yet their reliability for complex professional work remains unclear. We study this problem through Vibe Patenting, an end-to-end patent-drafting testbed for AI-agent evaluation. A separately-invoked LLM judge evaluates generated patent drafts and provides structured feedback for iterative revision. Across multiple inventions and drafting-agent configurations, judge-guided revision consistently improves judge-assessed quality, while unguided revision tends to saturate. Notably, iterative judge feedback enables a low-reasoning agent to approach the performance of a substantially more expensive high-reasoning agent. Stronger models and increased reasoning generally improve judge-assessed drafting quality, while domain-specific agentic workflows provide further gains. We validate the judge against independent evaluation by a professional patent attorney and find meaningful but strongly metric-dependent agreement and systematic calibration differences. These results highlight both the utility and limitations of LLM judges as evaluators and optimization signals for complex professional workflows.

---

145. Generalized Agent Iteration: One Formal Framework for Iterative Policy Improvement and Recursive Self-Improvement

• Authors: Hongyao Tang , Yi Ma , Pengyi Li , Yifu Yuan
• URL: https://arxiv.org/abs/2609.13406
• Abstract:

  When we speak of recursive self-improvement (RSI), are we speaking of a phenomenon, a mechanism, or a prospect? Towards autonomous and evolving intelligence, RSI is being claimed at many scales, while no single framework that formally describes these emerging instances exists. Its counterpart in the classical realm, iterative policy improvement, is characterized by generalized policy iteration (GPI), a framework of broad applicability with well-understood theoretical properties, but only where the update principle and the evaluation base lie outside the agent. In this paper, we propose Generalized Agent Iteration (GAI), a formal framework that describes iterative policy improvement and RSI as two cases of a single learning paradigm. GAI defines the agent as a configuration of modifiable components within a system and models the learning process as a cycle of agent evaluation and agent improvement. Two pivotal dials then distinguish the instances: whether the improving mechanism is part of the agent and whether the standard it is measured against is grounded outside it. The former dial delineates the boundary between GPI and RSI, and the latter determines a system’s polarity as anchored, goal drift, or fully self-referential. Moreover, we use these coordinates to place existing systems on the same two axes and make the defects of recursive self-improvement statable one condition at a time. We see this paper as a first step toward exploring a formal characterization of RSI that rests on the classical account, makes existing systems comparable, and provides a principled basis for analyzing and designing new ones.

---

146. Converge Then Diversify: Decoupling Convergence and Diversity in Multi-Objective Bayesian Optimisation

• Authors: Chao Jiang , Yueling Huang , Miqing Li
• URL: https://arxiv.org/abs/2609.13396
• Abstract:

  Multi-objective Bayesian optimisation (MOBO) is a sample-efficient approach for optimising expensive black-box functions with multiple objectives. In MOBO, the goal is to adequately approximate the Pareto front; that is, to obtain a high-quality solution set with 1) good convergence (closeness to the Pareto front) and 2) good diversity (spread across the Pareto front). Existing MOBO methods typically aim to accomplish these two tasks simultaneously, i.e., driving the search towards the Pareto front while maintaining a diverse set of nondominated solutions, such that the solutions, ideally, can gradually approach the entire front. When sufficient search budgets are available, this approach is effective. However, considering both convergence and diversity throughout the search is not easy and requires careful design. Under very tight budgets, there may not be enough solutions generated to be able to simultaneously approach the entire Pareto front. To address this issue, this paper proposes a \textit{converge-then-diversify} (CTD) approach that decouples convergence and diversity into two stages. In the first stage, CTD focuses on convergence, aiming to quickly drive the search toward a single point on the Pareto front. In the second stage, CTD focuses on diversity, aiming to spread solutions across the front. We present two simple instantiations of CTD by using widely adopted acquisition functions in the area. Experimental results show that, across all 446 pairwise comparisons, CTD statistically outperforms state-of-the-art methods in 72.9\% of the cases, performs equivalently in 21.1\%, and is statistically worse in only 6.1\%, with the advantage being particularly evident in settings with very tight evaluation budgets or in high-dimensional problems.

---

147. ZGCM-1: A Fully Open and Extremely Efficient Foundation Model for Math and Agentic Search

• Authors: Jiyan He , Guang Liang , Hao Liu , Haoxiang Guan , Jinbo Sun , Junyi Guo , Wenjun Feng , Yantai Xie , Yifei Shen , Bin Shao , Chuyang Wei , Kai Chen , Kexin Zhou , Minghang Zhu , Shuxin Zheng , Tie-Yan Liu , Taine Zhao , Wenhui Zhu , Xueyin Xu , Xiaoqing Zhang , Yatao Li , Yuxuan Ren
• URL: https://arxiv.org/abs/2609.13356
• Abstract:

  In this work, we present ZGCM-1, a fully open 7B dense foundation model trained from scratch with extreme data, system, and algorithmic efficiency. ZGCM-1 is founded on a core premise: compact models cannot passively memorize the open web, but can overcome parametric capacity limits by coupling deliberate internal thinking with active external tool use. To support this paradigm across a 256K context, we develop an end-to-end, high-efficiency open training recipe: Architecture & System Co-design: interleaved gated sliding-window and full attention, and a stable FP8 Muon optimizer; Progressive Curriculum & MDP Mid-Training: context scaling across 16K, 64K, and 256K, and the reformulation of interaction traces into Markov Decision Processes. Furthermore, we establish an AI-native R&D workflow where agent swarms autonomously manage cluster operations, data curation, and rapid diagnostic evaluation. Extensive evaluations show that ZGCM-1-7B is competitive across 7B model family on general benchmarks. On several challenging mathematical reasoning and agentic search suites, it remains competitive with frontier models orders of magnitude larger, such as Qwen3-235B-A22B and GLM-5.1. We also show that our pre-training design offers a ~4.2x efficiency improvement in 16K pre-training time-to-loss. Across the full development lifecycle, we distill eight actionable empirical findings-spanning architectural scaling, SFT quality pruning, long-context generalization, and agentic co-training dynamics. To facilitate community research, we open-source model weights from the pre-training, mid-training, and post-training stages, intermediate checkpoints, training code, per-stage data and data recipes, and W&B logs.

---

148. The Router Within: Eliciting Native Skill Routing from a Frozen LLM

• Authors: Ruishuo Chen , Xun Wang , Yu Chen , Zhuoran Li , Longbo Huang
• URL: https://arxiv.org/abs/2609.15982
• Abstract:

  Skills extend an LLM agent beyond its parametric knowledge, and the gain they promise rests on picking the right one. Deployed harnesses route by preloading every skill’s metadata into the context, which disperses the agent’s attention and caps the library size. Retrieval pipelines move the selection out of the context, but also out of the agent’s capability. We show that the frozen agent LLM already carries the routing signal in its own forward passes, and that two linear maps suffice to read it out with no skill text in the context. Gavel (Glance And Verdict from a frozen LLM) reads it in two steps. A glance projects the task’s and each skill’s mid-layer states through the two maps, the only parameters trained, and scores the full library against compact per-skill banks that one forward pass builds at installation. A verdict then resumes the shortlisted skills’ forward passes and reads the model’s own likelihood and yes/no judgment, fused with the glance as a product of experts. Trained once, Gavel transfers zero-shot to three public benchmarks and SkillTraj, our new benchmark of 372 simulated agent trajectories. On Qwen3-32B it outperforms progressive disclosure and retrieve-and-rerank pipelines that add 1.2B to 16B external parameters, by up to 13.4 points on written tasks and up to 21.9 when the need for a skill arises mid-rollout. Routing accuracy improves as the backbone does, and in a bash-agent harness the same 32B triggers the correct skill on Skill-Use more often than far larger frontier models running in Codex.

---

149. Vulnerability Localization Benchmark: Measuring Agentic Security Analysis at Repository Scale

• Authors: Aman Priyanshu , Supriti Vijay , Kimia Majd , Xuhong He , Fraser Burch , Takahiro Matsumoto , Jianliang He , Baturay Saglam , Arthur Goldblatt , Zhuoran Yang , Amin Karbasi
• URL: https://arxiv.org/abs/2609.15939
• Abstract:

  Language-model agents increasingly operate over complete software repositories, yet cybersecurity evaluations primarily measure whether they can detect, reproduce, or repair vulnerabilities rather than whether they can locate the relevant code. We study vulnerability localization: given a weakness class and an unfamiliar repository, identify the implementation files associated with that weakness. We introduce the Vulnerability Localization Benchmark (VLoc Bench), comprising 500 real world vulnerabilities from 290 repositories across six package ecosystems and 147 CWE categories. Each task pairs repository snapshots immediately before and after a security fix. On the vulnerable snapshot, an agent receives only the CWE description and read-only terminal access and must return the affected files; on the patched snapshot, it must determine that the recorded vulnerability is no longer present. We evaluate 27 language models and four static-analysis tools under a common agent interface. Repository-scale vulnerability localization remains difficult: the strongest system achieves 0.229 File F1, and 38.4% of tasks receive no correct localization from any evaluated model. We further find that stronger localization does not imply reliable behavior after remediation: systems that identify vulnerable files effectively can still report unsupported locations on patched repositories. These results establish vulnerability localization as a distinct repository-scale capability and provide a setting for studying both how security agents search for vulnerable code and when they should refrain from reporting it.

---

150. SlipSense: Multimodal Tactile Learning for Low-Latency and Generalized Slip Detection

• Authors: Tong Jian , Aditya Thurvas Senthil Kumar , Xinyi Li , Ziling Chen , Tianyu Dai , Ali Sengul , Matteo Grimaldi , Wenjie Lu , Saleh Nabi , Tao Yu
• URL: https://arxiv.org/abs/2609.15910
• Abstract:

  Slip detection is fundamental to dexterous manipulation, yet existing systems often lack precise characterization of detection latency and cross-platform generalization. We present SlipSense, a multimodal tactile slip-detection framework built on TacV5, a compact sensor integrating a $32 \times 32$ piezoresistive array operating at 240 Hz and a 3-axis MEMS accelerometer operating at 8 kHz. The piezoresistive array captures spatial pressure distributions, while the accelerometer captures friction-induced vibrations, providing complementary slip cues. The framework performs modality-specific encoding, intra-sensor fusion, and cross-modal attention with causal temporal prediction at 240 Hz. Experiments on a dataset of 1.4 million frames spanning 37 objects demonstrate the complementarity of the two modalities. SlipSense achieves 96.7% Macro F1 with a false-positive rate below 1.6%, detecting 76% of slip events within 23.1 ms. When trained solely on UMI data, SlipSense generalizes zero-shot to a Tesollo dexterous hand, transferring across unseen objects, distinct sensor units, and robotic platforms without retraining.

---

151. Anatomical Grounding and Leakage-Aware Multimodal Contrastive Learning for Alzheimer’s Disease Classification from Structural MRI

• Authors: Paul-Gabriel Nicolae , Irina Georgiana Mocanu
• URL: https://arxiv.org/abs/2609.15888
• Abstract:

  Deep networks trained on structural MRI for Alzheimer’s disease (AD) staging often reach reasonable accuracy while attending to anatomically irrelevant regions, and multimodal models that add clinical tables frequently rely on variables that were used to assign the diagnostic label in the first place. We study both issues with a deliberately lightweight slice-based encoder (ResNet18 with a one-layer Transformer over slices) on 1,075 baseline T1-weighted scans from ADNI-1. First, we use FastSurfer segmentations as an anatomical reference: YOLOv8 models trained on segmentation-derived labels localize Alzheimer-relevant structures with mAP_50 above 0.96, and a Grad-CAM comparison shows that the image-only classifier frequently attends to the skull, orbits and background. Second, we adapt a CLIP-style image - tabular contrastive framework and organize ADNIMERGE variables along a label-leakage spectrum. Fusion with cognitive scores yields 87.3% three-way accuracy, which we treat as a leakage-driven upper bound rather than an imaging result; fusion with regional volumes yields 73.0%. We observe that the choice of contrastive target changes what the image encoder learns: on MCI vs. CN, the image-only head reaches 52.4% when the encoder is aligned to cognitive scores and 73.8\% when aligned to volumes, although no tabular input is used at inference. Third, restricting the input to a per-subject crop of the medial temporal lobe raises image-only three-way accuracy from 58.7% to 65.1%. All results come from single runs on a small balanced test set, and we report confidence intervals and the protocol differences that prevent direct comparison with published numbers.

---

152. Privacy-enhanced federated learning via asynchronous aggregation and local differential perturbation

• Authors: Zhen Zhong (1), Shini Yang (2), Liesheng Wei (3) ((1) Georgetown University, Washington, D.C., USA, (2) LinkedIn, CA, USA, (3) Shanghai Ocean University, Shanghai, China)
• URL: https://arxiv.org/abs/2609.15885
• Abstract:

  This study proposes a privacy-enhanced federated learning framework to address secure collaborative training in distributed data environments. The framework integrates Dynamic Differential Privacy (DDP), lightweight Homomorphic Encryption (HE), and Local Differential Privacy (LDP) mechanisms to ensure data privacy protection during model training. Additionally, the framework employs an asynchronous aggregation strategy with version control to support distributed training in asynchronous environments. Experimental validation on the CIFAR-10 and Purchase-100 benchmark datasets demonstrates that the method maintains high classification accuracy (up to 82.6%) even under stringent privacy constraints ({\epsilon} = 0.1), while reducing communication overhead by 21.3% compared to FedAvg. Experimental results demonstrate that this framework effectively balances privacy protection and model performance in distributed machine learning scenarios, providing a scalable technical foundation for large-scale distributed collaborative computing.

---

153. Learning Multimodal One-step Flow Policy via Value-weighted Optimal Transport

• Authors: Jaehun Shon , Jinha Choi , Jongwook Jeon , Jongmin Lee
• URL: https://arxiv.org/abs/2609.15883
• Abstract:

  Offline reinforcement learning aims to learn a policy solely from fixed datasets, which often contain multimodal action distributions. Flow policies can naturally represent such multimodal behaviors, but learning an efficient one-step flow policy remains challenging: standard value guidance often leads to mode collapse or exploits overestimation bias in out-of-distribution regions. To address this, we introduce One-step Flow policy via Optimal Transport (OptiFlow), a framework for one-step flow policy learning as a structured sample-allocation problem. OptiFlow jointly trains a value-aware reference flow policy and an efficient one-step policy, coupling their action samples through state-wise entropic optimal transport. For each state, critic-estimated values define the priority of distillation target actions, while the action-distance cost ensures geometrically compatible pairings. By avoiding direct critic maximization, our transport-guided approach enables in-distribution exploitation by anchoring the one-step policy to high-value, dataset-supported modes without the risk of out-of-distribution divergence. Experimental results demonstrate that OptiFlow effectively captures optimal multimodal behaviors and achieves strong performance across diverse offline RL benchmarks. Our code is available at this https URL .

---

154. LLM-Based Schema-Aware Split Learning for Privacy-Preserving Mental Distress Prediction Across Heterogeneous Surveys

• Authors: Md Khalid Syfullah , Alvi Ataur Khalil
• URL: https://arxiv.org/abs/2609.15871
• Abstract:

  Rising societal and lifestyle complexity has been linked to a growing prevalence of mental distress worldwide. Educational institutions, workplaces, clinics, etc. collect large volumes of mental health survey data to understand and reduce this burden. Collaborative analysis of such data could yield effective generalizable predictive models. Privacy constraints and varied survey designs (i.e., different questions, scales, and formats) hinder direct integration. We propose a schema-aware split learning (SL) framework that preserves privacy, using a large language model (LLM) as a shared semantic encoder to harmonize heterogeneous survey schemas across institutions. We serialize each survey record into a natural-language description, unifying disparate survey schemas into a common format. The LLM is fine-tuned for mental distress assessment via Low-Rank Adaptation (LoRA) and partitioned across client and server. Clients retain the raw survey responses locally and run only a lightweight front-end, so original records never leave the institution that collected them. The resource-intensive backbone runs on the server, minimizing client-side computation. Using LLaMA-3.2-3B-Instruct, the framework attains an average ANLS of 0.708 with only 2,000 training samples, surpasses federated learning (FL) in eight of nine settings, and cuts per-client computation by three orders of magnitude, while generalizing to unseen datasets. Overall, it enables accurate, privacy-preserving, and resource-efficient collaborative learning from heterogeneous mental health survey data.

---

155. K-Bench: a clinically calibrated benchmark for evaluating large language models in high-risk mental health conversations

• Authors: Laura M. Vowels , Matthew J. Vowels , Shivali Sharma , Apoorv Jha , Rehnuma Choudhury , Wasseem El Sarraj , Rachel Francois-Walcott , Aruba Hussain , Sarah Ingram , Angela Loulopoulou , Adva Segal , Elena Volkova
• URL: https://arxiv.org/abs/2609.15855
• Abstract:

  % !TEX root = ../main.tex People increasingly use large language models (LLMs) for mental health support, yet their safety in evolving, high-risk conversations remains poorly characterised. We developed K-Bench, a clinician-calibrated, protected benchmark evaluating 125 model configurations representing 33 base models from 14 providers across a fixed cohort of 200 multi-turn vignettes involving suicide, self-harm, domestic violence, substance misuse, and no-risk presentations. Synthetic patient conversations showed substantial distributional overlap with real human-AI conversations. A frozen GPT-4o judge achieved 94.2% exact agreement with clinician consensus across 6,751 eligible item comparisons from 151 clinician-rated transcripts. Leading models combined strong supportive conversation with combined-risk scores above 95, whereas risk exploration exposed substantial variation among lower-performing configurations. Therapeutic prompting produced configuration-specific gains concentrated among weaker models, while elevated reasoning produced no average improvement. K-Bench combines broader clinical coverage and configuration-scale comparison with a continuously updated public leaderboard whose operational test materials are protected from direct optimisation. The leaderboard is available at this http URL .

---

156. Before You Poll with LLMs: A Deliberative Diagnostic Framework

• Authors: Ahmed Wali , Hassaan Tayyab
• URL: https://arxiv.org/abs/2609.15849
• Abstract:

  Can LLMs reason through new information like humans, or do they merely retrieve cached opinions? This is critical for silicon sampling, where LLM personas simulate public opinion at scale. Current evaluations test only whether personas hold the right opinions – a static snapshot. But opinion research increasingly depends on dynamic fidelity: whether personas update beliefs in response to new arguments, as humans do during deliberation. No existing benchmark tests this. We introduce the Deliberative Polling Diagnostic Framework, which compares human and LLM belief shifts after identical informational interventions. Grounded in deliberative polling, it surfaces failures invisible to static evaluation: models that produce plausible partisan opinions can still misrepresent how those opinions change. Applying the framework to five frontier models using data from America in One Room (526 personas, 72 questions), we find that every model fails, each in a unique manner. GPT-5.1 exhibits reversal: its personas become more hostile toward the opposing party after balanced information, while humans become less so. This reversal is selective (80% on outgroup vs. 26% on policy questions) and symmetric across partisan identities. Gemini 2.0 Flash, Claude Sonnet 4.5, and Llama 3.3 70B exhibit overshoot, shifting correctly but at 5-7x human magnitude. DeepSeek V3 exhibits rigidity with near-zero change. Targeted ablations reveal that policy content triggers these failures and that they are identity-specific: GPT-5.1 reverses on outgroup questions but overshoots on ingroup; Gemini shows the inverse. We term this signature self-sycophancy: conformity to the model’s internal stereotype of the persona rather than reasoning from the information provided. Our framework offers a concrete protocol: run the deliberative diagnostic before trusting LLM personas to mimic revised beliefs.

---

157. Per-Matrix Optimality Is Not Enough: Three-Level Optimization for Low-Rank LLM Compression

• Authors: Huicheng Zhang , Xiyao Feng , Ze-Tong Li , Chengkai Zhu , Xiao Shi , Xiwei Pan , Jinguo Liu , Ge Bai , Xin Wang
• URL: https://arxiv.org/abs/2609.15838
• Abstract:

  Per-matrix singular value decomposition (SVD) truncation is Eckart-Young optimal in the whitened Frobenius norm, but errors from independently compressed matrices compound through the block’s nonlinear forward pass. Inspired in part by hierarchical variational optimization in quantum many-body methods, we introduce a three-level chain that widens optimization scope from individual matrices to Transformer blocks to the full model: whitened SVD~(L1), block-level joint optimization~(L2), and end-to-end language-modeling loss refinement~(L3), all from 256 calibration sequences, with no instruction or recovery data. On LLaMA-7B at 60% compression, the chain reduces WikiText-2 perplexity from 42.1 to 19.1 to 11.4. The block-level stage acts as a regularizer: skipping it worsens Penn Treebank (PTB) perplexity by 24 points, a gap that additional end-to-end training did not close in our experiments. Perplexity gains hold across 20-80% compression, five architectures up to 13B parameters, and both in-distribution and out-of-distribution benchmarks, though the cross-architecture rows use architecture-specific configurations and the ratio sweep was not run under one common protocol. With more calibration data, skipping the block-level stage becomes competitive, revealing an offline compute–data trade-off. We therefore claim improvements only in perplexity and compression fidelity; downstream accuracy remains well below the dense model.

---

158. CiteGuard-RAG: A Validation-Centered AI System for Evidence-Grounded Question Answering

• Authors: Sumit Barua , Guan Hong , Halil Dursunoglu , Charles Rodgers , Alvis Fong
• URL: https://arxiv.org/abs/2609.15830
• Abstract:

  Retrieval-augmented generation (RAG) can improve access to complex information; however, retrieving evidence alone does not ensure that answers are grounded, citation-valid, or appropriately refused. This paper introduces CiteGuard-RAG, a validation-centered AI system for evidence-grounded question answering. The system integrates hybrid semantic-lexical retrieval, citation-constrained generation, sentence-level grounding validation, and single-pass regeneration. Validation is used at runtime to determine whether a candidate answer should be accepted, refused, or regenerated before final delivery. CiteGuard-RAG is evaluated on 400 questions across a controlled housing-law dataset, PrivacyQA, and CUAD. In the controlled evaluation, it achieves 99.1% retrieval accuracy, 98.3% grounded-answer accuracy, and 98.3% citation validity, with no validation-detected hallucinations. Ablation results show that grounded-answer accuracy drops sharply when validation is removed, even when retrieval accuracy remains unchanged. External evaluation shows that while citation validity remains strong, evidence utilization, span alignment, and refusal calibration become harder under domain shift. These findings indicate that trustworthy RAG systems require explicit validation between retrieval and final answer delivery. CiteGuard-RAG provides a practical architecture for linking retrieval, generation, citation checking, abstention, and regeneration in high-stakes information access.

---

159. Delegating Authorization to Misaligned Agents: Coalitional Alignment and Safe Control

• Authors: Natalie Collina , Surbhi Goel , Aaron Roth , Sikata Bela Sengupta
• URL: https://arxiv.org/abs/2609.15803
• Abstract:

  Long-running AI agents create a control problem: each action they take changes the state, which in turn affects the trajectory of future actions. If the agent is not fully aligned, then guaranteeing safety requires approving consequential actions before allowing them to be executed. But requiring human approval at every step makes attention a bottleneck. Delegating review to other AI agents raises the same alignment problem: the reviewers may themselves be misaligned. We identify a condition on a reviewing panel that is weaker than individual alignment yet necessary and sufficient for a guarantee that the principal fares at least as well in expectation as under a designated baseline policy. Each reviewer agent reports whether an action proposal made by a proposer agent improves its own utility relative to the baseline. We show that a threshold rule tolerating $k$ disapprovals is safe exactly when, after any $k$ reviewers are removed, the principal’s utility can be written as a nonnegative combination of the remaining reviewers’ utilities, plus a term that is nonnegative on every feasible proposal. We call this property $k$-robust coalitional alignment. The characterization lifts to sequential control: in a discounted MDP with an arbitrary proposer agent, safety at every state is both necessary and sufficient for the induced policy to match or improve on the baseline. When reviewers vote strategically, full-panel coverage in reward-function space guarantees that every Nash equilibrium is safe under the unanimous approval rule; in contrast, more permissive thresholds can admit unsafe equilibria even when reviewers are individually aligned. Experiments with existing reviewer models show that collective review can remain sound without an aligned individual, even when some disapprovals are tolerated.

---

160. When the World Lies: Backdoor Attacks on Latent World Models for Downstream Control

• Authors: Roberto Riaño , Gorka Abad , Stjepan Picek , Aitor Urbieta
• URL: https://arxiv.org/abs/2609.15781
• Abstract:

  Pretrained world models, learned simulators that encode an observation into a latent state and predict how it evolves under actions, are beginning to be reused as off-the-shelf dynamics backbones for control, like pretrained encoders and language models are reused today. We show that this reuse opens a supply-chain backdoor: an adversary who controls only a released checkpoint can hijack the downstream controller, even though the victim trains and evaluates entirely on clean data and never sees the trigger. The attack encodes no explicit trigger-to-action rule. Instead, the poisoned model routes trigger-bearing observations into a chosen latent region and reshapes the local dynamics there, so that the victim’s own optimization (Dreamer-style actor training in imagination, or MPC/CEM planning over predicted futures) re-discovers the attacker’s target action on its own. Across several control tasks and trigger families, the trigger steers the controller’s action toward the attacker’s target, controlling every action dimension and hijacking 100\% of triggered steps on the strongest settings. The checkpoint still passes the clean-data diagnostics a victim would run before deployment, with clean-task success retaining at least $\sim$75\%. The effect is temporally gated: it appears only while the trigger is present and disappears when the trigger is removed. Trigger-blind repair is budget-dependent: moderate clean fine-tuning can preserve clean utility while leaving the triggered failure intact, whereas sufficiently aggressive adaptation can remove it only after substantially degrading clean control. The world-model backbone itself is therefore an emerging and underexamined attack surface for control. The full code and artifacts are available in our repository.

---

161. Transfer Learning for Socioeconomic Estimation in Forced-Displacement Settings

• Authors: Steven Ndung’u , Adel Daoud , Ismael Yacoubou Djima , Hai-Anh H. Dang , Patrick Michael Brock
• URL: https://arxiv.org/abs/2609.15773
• Abstract:

  Progress in inclusive household surveys has strengthened socioeconomic evidence for forcibly displaced populations, providing indispensable benchmarks on living conditions and welfare. However, these surveys remain resource-intensive and periodic, while conditions can change between rounds, particularly in settings affected by fragility, conflict, and violence. More frequently updated, spatially granular complementary evidence is therefore needed to identify where socioeconomic conditions may be changing between survey rounds and to inform operational prioritization. Earth observation and machine learning offer a scalable source of spatially explicit socioeconomic information. However, tools developed for general populations have not been systematically adapted and evaluated in forced displacement settings, where living conditions, settlement patterns, and displacement impacts may differ substantially. We address this gap by adapting a multimodal spatiotemporal vision transformer, pretrained on Demographic and Health Survey data from approximately 1.2 million households across 36 African countries, to forced displacement and host community settings in South Sudan, Cameroon, and Zambia. We develop and evaluate the updated, adapted model using socioeconomic indices derived from UNHCR FDS and RMS data. Our results show that satellite-derived geospatial covariates explain up to 66% of the variation in socioeconomic outcomes in camp-intersecting grids, with a mean absolute error (MAE) of 4.37 index points, and 41% in non-camp-intersecting areas, with an MAE of 5.41. The framework complements and adds value to periodic household surveys by filling critical spatial and temporal data gaps with regularly updated, model-based socioeconomic estimates. These estimates sustain insight between survey rounds and support timely humanitarian prioritization and field verification.

---

162. Event-Native Symbolic-Temporal Spike Encoding Framework for Heterogeneous Cyber Streams

• Authors: Dalton Diez , Peyton Andras , Max Shroyer , James Ghawaly Jr
• URL: https://arxiv.org/abs/2609.15772
• Abstract:

  Spiking neural networks (SNNs) have shown promise for sparse, event-driven computation through stateful processing that is naturally compatible with low-power edge hardware. These properties align with cyber monitoring, where data arrives asynchronously, and malicious behavior often emerges through temporal patterns across event sequences. However, cyber streams are not composed solely of continuous numeric signals: their informative structure is also carried by categorical identifiers, irregular timing, and local behavioral context. Traditional rate- and population-based spike encodings are not naturally suited to these heterogeneous semantics, while conventional intrusion detection system (IDS) pipelines typically resolve the mismatch by converting raw events into flows, fixed aggregation windows, or dense tensors. Although useful for conventional classifiers, these transformations introduce buffering latency, obscure native temporal structure, and weaken the computational advantages of event-driven neuromorphic processing. We introduce an event-native symbolic-temporal spike encoding framework that maps heterogeneous cyber events directly into sparse, spike-compatible inputs. By assigning encoding roles to semantic identity, local frequency context, and inter-event timing, the framework preserves categorical semantics and temporal dynamics. We validate the approach on packet-level Network IDS and extend it to message-level CAN IDS, using both domains to evaluate whether the encoding exposes usable structure for recurrent SNNs operating directly on native event streams. Under edge-oriented, $\mu$Caspian-aligned hardware constraints, compact recurrent SNNs achieve strong anomaly detection performance, with an operational hybrid metric ($J_{hybrid}$) of 0.987 on Network IDS and 0.980 on CAN IDS.

---

163. Sylvas: Synergistic Learning Value based Device Scheduling in Federated Continual Learning

• Authors: Yuxuan Sun , Yuxuan Bai , Tan Chen , Sheng Zhou , Zhisheng Niu
• URL: https://arxiv.org/abs/2609.15763
• Abstract:

  Federated continual learning (FCL) enables shared global models to continuously adapt to distributed and non-stationary data streams, making it important for Internet of Things applications such as intelligent transportation, industrial monitoring, and unmanned systems. Under spatio-temporal data distribution dynamics and label scarcity, a key challenge is how to quantify the contribution of each edge device to global learning performance and schedule the most valuable devices under resource constraints for timely model updating. This article presents Sylvas, a synergistic learning value based device scheduling framework for FCL at the wireless edge. Sylvas evaluates the learning value of distributed data from two perspectives: distributional value, which characterizes the contribution of device data to global model learning from a spatio-temporal distribution perspective, and label value, which captures the quantity and reliability tradeoff of pseudo-labeled data. By integrating these factors into a synergistic learning value metric, Sylvas schedules devices with high learning value while satisfying communication and computation resource constraints. Case studies demonstrate that Sylvas supports timely model adaptation under spatio-temporal distribution dynamics and effectively exploits unlabeled data.

---

164. Look Before You Leap: Factual Decoding with Internal Attribution Signals

• Authors: Hayeong Ryu , JungMin Yun , Byeonggeuk Lim , Sunhee Jo , YoungBin Kim
• URL: https://arxiv.org/abs/2609.15745
• Abstract:

  Hallucination remains a critical challenge in large language models (LLMs), where early factual errors compound through autoregressive generation in a snowballing effect that neither post-hoc correction nor weight-level intervention can effectively preempt. We propose DescaPE (DEcoding Signal Control Against Path Error-snowballing), a decoding framework that leverages internal model signals to suppress hallucination-prone trajectories at inference time. Through sliding-window MLP ablation, we identify a factual-salient layer span within LLMs whose derived signal is selectively elevated for factual tokens and exhibits anomalous spikes at hallucination-prone steps. We train a lightweight probe to approximate this signal from a single forward pass and integrate it into candidate scoring to penalize high-risk continuations while rewarding factually grounded ones. Experiments across five factuality benchmarks on three LLMs demonstrate that DescaPE achieves factuality improvements over decoding-time baselines in multiple settings, while incurring only 1.10x latency overhead in our efficiency evaluation. Our code is available at this https URL .

---

165. A Language-Guided Multimodal Foundation Model for Zero-Shot and Multi-Task Brain Signal Analysis

• Authors: Mingzhi Chen , Yiyu Gui , Guibo Luo , Yuchao Yang
• URL: https://arxiv.org/abs/2609.15740
• Abstract:

  Brain signal analysis is essential for both neuroscience research and clinical diagnostics, yet current approaches face critical limitations. End-to-end models require task-specific retraining and exhibit limited generalization, while pre-trained models lack semantic depth and still depend on extensive fine-tuning. Meanwhile, general-purpose multimodal foundation models, though powerful in other domains, struggle to interpret brain signals due to representational misalignment and lack of domain knowledge. This study introduces a multimodal foundation model for zero-shot and multi-task brain signal analysis (METIS) through a unified language-signal alignment framework. METIS is pretrained on the largest and most diverse brain-signal corpus to date, comprising over 70,000 h of recordings from more than 11,000 subjects across 20 datasets. In a comprehensive zero-shot evaluation across 12 datasets, METIS outperformed the leading generalist model by over 20.9% in average accuracy. Remarkably, without any fine-tuning, METIS’s performance matches or exceeds that of supervised, task-specific models. Furthermore, METIS demonstrates exceptional data efficiency and strong generalization, achieving an average AUROC advantage of over 16.0% in few-shot settings and 15.9% in cross-dataset transfer. This work establishes a new paradigm for general-purpose brain signal analysis, paving the way for next-generation neurotechnology.

---

166. Bench2Dex: Benchmarking Visuo-Tactile Bimanual Dexterous Manipulation Across Dexterous Hands

• Authors: Zhenjie Yang , Yideng Zhang , Dongjie Zhang , Chenyu Jiang , Xianshuai Liu , Yufeng Li , Zuhao Ge , Xingyu Jiao , Zheng Zhang , Kaiyu He , He Wang , Yuwen Zhong , Yi Deng , Muyun Jiang , Xianliang Huang , Haisheng Su , Donghang Zhang , Jian Zhang , Xue Yang , Hongyang Li , Zuxuan Wu , Yu-Gang Jiang , Xiaosong Jia , Junchi Yan
• URL: https://arxiv.org/abs/2609.15726
• Abstract:

  Tactile sensing provides contact information that can be difficult to infer from vision alone, but tactile hardware for dexterous hands has not converged to a common design. Dexterous hands differ in finger structure, contact surfaces, and sensor layouts, while simulated tactile signals still differ from measurements produced by physical sensors. These factors make it difficult to study visuo-tactile manipulation across diverse dexterous hands within a consistent experimental setting. We present Bench2Dex, a simulation benchmark for visuo-tactile bimanual manipulation across 12 dexterous hands. We adapt existing robot models with a shared simulated tactile interface that converts local contact geometry into image-like tactile observations. The interface provides a consistent observation format across different hand morphologies without attempting to reproduce the output of a specific physical tactile sensor. Bench2Dex includes 26 bimanual manipulation tasks that involve tool use, articulated-object interaction, and multi-stage manipulation, together with about 1.3K human-teleoperated demonstrations. The benchmark provides synchronized visual, tactile, proprioceptive, action, and object-state observations, together with executable task metrics. For robustness, we group seven perturbation types into invariance axis, where the correct action does not change, and equivariance axis, where the correct action changes together with the perturbation. We evaluate ACT, Diffusion Policy, pi0.5, and GR00T N1.5 on Bench2Dex and report their performance and failure modes. Bench2Dex is meant as a platform for studying visuo-tactile learning across dexterous hands. It does not assume that simulated tactile observations can replace real tactile sensing; it offers a shared setting for algorithm development while tactile hardware and simulation models are still evolving.

---

167. More Than Just Access: Generative AI as Communication Intermediary for Blind and Low-Vision Users

• Authors: Protik Dey , Mohd Saifuzzaman , Taslima Akter
• URL: https://arxiv.org/abs/2609.15696
• Abstract:

  Generative AI (GenAI) tools are increasingly woven into how blind and low-vision (BLV) people communicate, not only with digital information, but with the physical world and with other people. Tools such as ChatGPT, Google Gemini, Be My AI, and Seeing AI translate visual and textual content into accessible form, and are beginning to substitute for interpersonal requests for help, such as asking a family member to read a label or describe a scene. Drawing on semi-structured interviews with 19 BLV participants, we examine GenAI as a communication intermediary and how it succeeds and fails as an alternative for reading, describing, and even asking another person for help. We also investigated what BLV users gain and risk when these tools take over that role. We conclude with design and policy implications for GenAI systems that communicate uncertainty honestly, protect information, and support BLV users’ independence rather than substitute for it unsafely.

---

168. Scalability and Performance Evaluation of Federated Learning Frameworks: A Comparative Analysis

• Authors: Bassel Soudan , Sohail Abbas , Ahmed Kubba , Manar Wasif Abu Talib , Qassim Nasir
• URL: https://arxiv.org/abs/2609.15681
• Abstract:

  This paper presents a systematic examination and experimental comparison of the prominent Federated Learning (FL) frameworks FedML, Flower, Substra, and OpenFL. The frameworks are evaluated experimentally by implementing Federated Learning over a varying number of clients, emphasizing a thorough analysis of scalability and key performance metrics. The study assesses the impact of increasing client counts on total training time, loss and accuracy values, and CPU and RAM usage. Results indicate distinct performance characteristics among the frameworks, with Flower displaying an unusually high loss, FedML achieving a notably low accuracy range of 66% to 79%, and Substra demonstrating good resource efficiency, albeit with an exponential growth in total training time. Notably, OpenFL emerges as the most scalable platform, demonstrating consistent accuracy, loss, and training time across different client counts. OpenFL’s stable CPU and RAM underscore its reliability in real-world scenarios. This comprehensive analysis provides valuable insights into the relative performance of FL frameworks, offering good understanding of their capabilities and providing guidance for their effective deployment across diverse user bases.

---

169. Don’t Send What You Don’t Need: Question-Guided Token Pruning as a Privacy Defense for Vision-Language Models

• Authors: Md Khalid Syfullah , Alvi Ataur Khalil
• URL: https://arxiv.org/abs/2609.15671
• Abstract:

  Visual Question Answering (VQA) with Vision-Language Models (VLMs) is increasingly used in privacy-sensitive and bandwidth-constrained settings. Federated Learning (FL), Split Learning (SL), and U-Shaped Split Learning (USL) keep raw data local, but transmitting all visual tokens across a model partition remains costly and can expose private information. We propose QPriv-VL, a question-guided, privacy-aware token-pruning framework for FL, SL, and USL that prunes visual tokens before transmission based on task utility and privacy sensitivity. Its core component is a lightweight Dynamic Threshold Predictor (DTP) that jointly estimates a sample-specific pruning ratio and a token-level retention mask in one forward pass. DTP combines question relevance, computed from cross-modal similarity between visual patches and the pooled question embedding, with a sensitivity signal derived from frozen DINOv2 features. This allows the model to suppress potentially sensitive regions while preserving patches useful for answering the question, without requiring sensitivity labels. We evaluate QPriv-VL on GQA, OK-VQA, VQAv2, SLAKE, VQA-RAD, and PathVQA against four privacy attack families: FSHA, FORA, iDLG, and attribute-inference membership inference attacks. DTP matches or outperforms fixed-ratio pruning while using substantially fewer transmitted tokens. On VQA-RAD, it reduces membership-inference attack success from 0.99 to 0.76-0.79, lowers FSHA and FORA reconstruction PSNR relative to fixed-ratio pruning, and preserves competitive VQA accuracy using about 40% of the original visual-token budget. A sensitivity exclusion ratio of 1.20 +/- 0.18 indicates preferential removal of privacy-sensitive patches, while explainability analysis shows that retention adapts to question semantics rather than generic visual saliency.

---

170. Benchmarking Intra-Patient 3D Deformable Multimodal Image Registration

• Authors: Matteo Barbieri , Giammarco La Barbera , Juan Pablo De La Plata , Sabine Sarnacki , Isabelle Bloch , Pietro Gori
• URL: https://arxiv.org/abs/2609.15669
• Abstract:

  Multimodal image registration is a key component of many clinical workflows, yet it remains challenging because corresponding anatomical structures often exhibit substantially different image intensities across modalities. In this work, we present a comprehensive benchmark of intra-patient 3D multimodal deformable registration methods across three datasets covering different anatomical regions and difficulty levels, including both synthetic deformation recovery and real clinical scenarios. We evaluate classical optimization-based approaches and modern learning-based methods, including recent deep learning and foundation models, using complementary metrics: Average Dice similarity coefficient (DSC), average 95th-percentile Hausdorff distance (HD95), and a modality-independent structural similarity measure based on the MIND self-similarity context (MIND-SSC). Results show high variability across datasets, with learning-based methods demonstrating superior performance on large synthetic benchmarks, while only limited improvements are observed in real pelvic registration. A key finding of this study is the consistent disagreement between geometric metrics (DSC, HD95) and image-based similarity metrics (MIND-SSC), highlighting that improved overlap does not necessarily imply better global multimodal correspondence. Furthermore, anatomy-guided approaches achieve the highest overlap scores but exhibit degraded performance outside of segmented regions, revealing a trade-off between label-driven alignment and global structural coherence. Overall, our results indicate that no current method achieves robust performance across anatomies and modalities. We demonstrate that intra-patient 3D multimodal registration requires multi-criteria evaluation, including deformation-based metrics, and remains an open problem.

---

171. Circuit-MLLM: Topological Logic-Guided Latent-Space Visual Reasoning for Circuit Schematic Understanding

• Authors: Jinyuan Deng , Yuqi Jiang , Wenjing Huang , Xin Li , Qi Sun , Cheng Zhuo
• URL: https://arxiv.org/abs/2609.15668
• Abstract:

  Through pre-training on extensive text and image datasets, current multi-modal large language models (MLLMs) achieve strong performance on general tasks. However, circuit schematics present a unique challenge for MLLMs due to their dense component layouts and distinct topological logic, demanding fine-grained structural parsing to extract the electrical semantics. To address this, we propose Circuit-MLLM, a multimodal reasoning framework that reformulates circuit topology analysis as a process of device localization, path tracing, and sequential reasoning within the latent space. We introduce a circuit knowledge mining mechanism that deeply aligns the model’s latent representations with structurally rich features derived from multi-granularity circuit vision experts, enabling the model to effectively internalize topological semantics. Building upon these internalized semantics, we devise a topology-guided sequencing strategy that decouples reasoning from the rigid raster-scan order, enforcing stepwise inference along the circuit’s topological logic in latent space. Across diverse circuit analysis tasks, Circuit-MLLM consistently outperforms strong baselines, notably achieving a 25% higher average score than GPT-5.1, which demonstrates the effectiveness of our framework in circuit schematic topology analysis. Code is publicly available at this https URL .

---

172. CIDERS: Cloud-Edge LLM Collaborative Learning via Accelerating Personalized Bilevel Optimization

• Authors: Victor H. Chen , Hairui Yu , Stella K. Chung , Hong Yan
• URL: https://arxiv.org/abs/2609.15664
• Abstract:

  Amid the rapid advancement of physical-world intelligence, cloud-edge collaborative large language models (LLMs) have emerged as a promising roadmap for practical LLM deployment. However, existing cloud-edge paradigms struggle to balance global consensus with local personalization, which fails to satisfy the need for a unified knowledge foundation on the cloud and domain-specific adaptation at the edge. To address this, we introduce, for the first time, a personalized bilevel optimization framework that formalizes cloud-edge LLM collaboration as a dual structure: the upper level optimizes edge-side personalization, while the lower level governs cloud-side knowledge transfer, reaching cloud-edge evolving in coordination. We then propose CIDERS, an efficient solver that decomposes the model into a learnable backbone and a messenger. While the cloud performs knowledge transfer to the learnable backbone, the key lies in embedding global trajectories into each local personalization step via consensus-variate correction to reconcile personalization with consensus. We provide a comprehensive theoretical analysis, including a geometric characterization of the local trajectory and a full convergence guarantee, revealing an explicit trade-off structure between personalization and global convergence. Extensive experiments demonstrate that CIDERS consistently outperforms competitive baselines on the compressed edge path, with 3.1x and 1.7x gains on mathematical reasoning and code generation, respectively, and a 10\% relative gain on instruction metrics. Mechanism experiments attribute these gains to early consensus-corrected coordination and task-aware distillation. Overall, CIDERS offers a viable path toward consensus-guided continuous personalization in cloud-edge LLM systems.

---

173. Kaininja: Extending Native 3D Generators to the Part Level

• Authors: Ruihan Yu , Lian Fu , Muyao Niu , Zheng-hui Huang , Yu-Ju Tsai , Sho Kuno , Fengbo Lan , Yonghao Yu , Erwin Wu , Ming-Hsuan Yang , Kaipeng Zhang , Zhixiang Wang
• URL: https://arxiv.org/abs/2609.15659
• Abstract:

  Native 3D generators turn one image into a single mesh. TRELLIS.2 and its peers deliver high-fidelity non-watertight geometry with materials, but the output is one fused object, while downstream work such as editing, rigging and simulation operates on part-level assets. A naive idea is to run a 3D segmentation network on the fused mesh that TRELLIS.2 generates, but such pipelines are slow and bounded by the accuracy of the segmentation. We want a simple way to extend an existing native 3D generator to the part level. But we face a critical problem: the O-Voxel grid stores one sheet of surface per voxel, so a single volume cannot represent the interface where two parts touch, at any resolution. We introduce a dual-volume representation to solve this problem and put forward KaiNinja, a part-level extension of TRELLIS.2 built on a dual-volume form of its O-Voxel representation. KaiNinja keeps the generation speed and quality of TRELLIS.2 while extending it to the part level, with no mask or segmenter in the pipeline. Its training data come from sources of many kinds, including CAD models and assets authored by an LLM-driven agent; to our knowledge it is the first 3D generative model trained on agent-authored part data. Surprisingly, we also find that whole-object fidelity improves over the same backbone fine-tuned on the same dataset. Against part generation pipelines of different paradigms, it lowers whole-object Chamfer distance by 40% and raises strict part F-score by 16%.

---

174. Predictive Likelihood Ratios for Language Model Watermark Detection

• Authors: Li Ma
• URL: https://arxiv.org/abs/2609.15657
• Abstract:

  Keyed watermark detection tests dependence between observed tokens and pseudorandom variables reconstructed from a secret key. Building on the pivotal framework of Li et al. (2025), we construct predictive likelihood ratios that average over uncertain probability deficits and residual-tail distributions. The aim is robust detection power across alternative specifications without requiring a single signal-strength tuning. A mixture prior combines tail shape and effective width; hierarchical extensions allow within-document variation in deficit or width. The test maximizes prior-averaged power at a fixed size, but is not generally uniformly most powerful or minimax. Under the exact conditional pivot null, normalized predictive alternatives selected before each observation yield a Bayes factor that is also a test martingale: Type I error control is unaffected by alternative misspecification and remains valid under optional stopping. This guarantee does not cover violations of the conditional null, and the interpolated implementation has no certified anytime guarantee. Gumbel marginal likelihoods are evaluated by fixed quadrature. Across the evaluated tail-shape and tail-width alternatives and three horizons, the union-tail mixture has maximum observed Type II error regret .0080, compared with .0962 for the equal-tail mixture, relative to the best tested rule. On temperature-matched outputs from two open models, it improves AUC over the equal-tail baseline in all eight non-saturated model-temperature cells, although the leading reference score generally has higher AUC. Supplementary experiments show retained power under independent null-like replacement and smaller changes from hierarchical dependence modeling. The evidence supports robustness across the evaluated alternatives, not uniform power guarantees or resistance to arbitrary text edits.

---

175. ModaLens: Measuring Image Sensitivity in Report-Conditioned Medical VLMs

• Authors: Sebastián Andrés Cajas Ordóñez , Maximin Lange , Quang Bui , Anqi Peter Li , Felipe Ocampo Osorio , Rafi Al Attrach , Kushul Reddy Palakala , Sahil Kapadia , Zakaria Laouabdia Sellami , Xinyue Zhang , Ashley Zhang , Leo Anthony Celi
• URL: https://arxiv.org/abs/2609.15635
• Abstract:

  A radiology report can already answer a clinical question, so it is hard to tell whether a vision-language model also uses the image. ModaLens, a paired image-swap audit, measures how report availability changes image sensitivity: MedGemma-27B on 3,199 paired MIMIC-CXR cases from 293 patients, all 14 questions per case (13 finding-specific and one composite), each image replaced by one from another study, usually of the same patient, with question and report fixed. Under an explicit answer instruction, the model’s generated answer changes on 4.26 percent of trials with the report and 20.94 percent without it, a paired increase of 16.7 points (patient-clustered 95 percent CI 15.6 to 17.7), so report availability reduces image-swap sensitivity under this protocol; the original prompt with a lowercase first-token readout gives 4.70 percent against 17.07 percent, and substitutions also move continuous answer scores where the binary prediction does not change. The labels are derived from reports, which limits conclusions about visual correctness; the direction replicates in two further model lineages. Code, the exact prompts and a run record for every number are at this https URL .

---

176. FedLTLib: A Comprehensive Benchmark for Federated Long-Tail Learning

• Authors: Changkun Lin , Junxiao Wang
• URL: https://arxiv.org/abs/2609.15625
• Abstract:

  Driven by the escalating demand for privacy-preserving computing, Federated Learning (FL) has witnessed remarkable progress, becoming a cornerstone technology for bridging distributed data silos in mobile edge networks. However, in real-world mobile computing environments, data is generated by heterogeneous mobile devices with varying user behaviors, leading to a significant Long-Tail Distribution. Unlike idealized balanced datasets, data in the wild manifests an acute imbalance where a minority of head classes dominate the sample space while a vast number of tail classes, often representing rare but critical edge-case events, are extremely scarce. This data heterogeneity, which we formally characterize as “Double Heterogeneity”, referring to the superposition of global class imbalance and local statistical skew, precipitates severe performance deterioration on tail classes, thereby spurring the vital research direction of Federated Long-Tail Learning (FL-LT). To standardize evaluation and accelerate research in this field, we introduce FedLTLib, a comprehensive benchmark tailored for FL-LT. Addressing the critical issues of inconsistent experimental configurations and unfair comparisons in prior work, FedLTLib establishes a standardized evaluation framework. The platform not only incorporates diverse benchmark datasets reflecting mobile data characteristics but also implements 13 state-of-the-art FL algorithms (4 traditional FL algorithms and 9 FL-LT algorithms). By leveraging FedLTLib, researchers can perform fair and reproducible evaluations of algorithm robustness and generalization capabilities under a unified experimental protocol, ultimately advancing the deployment of robust intelligence in mobile computing ecosystems.

---

177. Beyond AI Literacy: A Structured Review and Exploratory Meta-Analysis of Measures for Competent Generative-AI Use

• Authors: Daniele Veri’
• URL: https://arxiv.org/abs/2609.15624
• Abstract:

  Researchers assessing competent generative-AI use at work must choose among self-reports, objective tests, and measures of oversight and reliance. We conducted a structured, seeded review of 24 focal empirical publications, starting from the 2024 COSMIN-based review and adding a targeted update through 17 August 2026. We grouped the measures into four domains: knowledge and use, epistemic oversight, reliance calibration, and operational control of tool-using agents. In an exploratory meta-analysis, we pooled three direct subjective-objective correlations from one research program (REML r = .055; Hartung-Knapp 95% CI [-.047, .156]; combined reported N = 2,765). We could not resolve a discrepancy between the largest study’s reported correlation and p-value, leaving its weight uncertain. Adding a synthetic mean of 12 cross-factor correlations from a fourth study gave r = .079 (95% CI [-.025, .181]). This sensitivity analysis concerns a broader comparison. From this small evidence base, we cannot establish a population correlation, validate workplace cutoffs, or justify substituting self-ratings for performance scores. We identified tests of foundation knowledge (AICOS-S and GLAT) and measures of verification, reliance, trust, and dependency. We found no validated individual-level instrument in the focal corpus that tests the full combination of agent scope, permissions, recovery, state isolation, independent review, and evidence-based closure; some cover subsets. We propose a four-layer workplace battery with non-compensatory decision rules, but have not tested its thresholds or whether it improves on other assessment approaches.

---

178. Multi-View Molecular Representation Learning with Hierarchical Graphs and Contextualized Fingerprints

• Authors: Gwang-Hyeon Yun , Jong-Hoon Park , Bing Hu , Helen Chen , Anita Layton , Young-Rae Cho
• URL: https://arxiv.org/abs/2609.15611
• Abstract:

  Molecular property prediction requires representations that generalize from limited labeled data to structurally novel compounds. Existing molecular pretraining methods often rely on a single view: graph-based approaches model atom-bond topology but provide limited fragment-level supervision, whereas fingerprint descriptors encode chemical patterns but are typically used as fixed auxiliary features. We propose HiFi-Mol, a multi-view framework that separately pretrains a hierarchical graph encoder and a contextualized fingerprint encoder before downstream integration. The graph branch uses fragment-aware masking with multi-resolution supervision to capture substructure-aware representations, while the fingerprint branch tokenizes active entries from seven fingerprint families and applies masked language modeling to learn contextualized embeddings. During fine-tuning, HiFi-Mol combines projected multi-resolution graph features with fingerprint embeddings for downstream prediction. Evaluated on MoleculeNet benchmarks under the scaffold split, HiFi-Mol achieves a 2.77% improvement in average ROC-AUC over the best baseline across eight classification tasks while maintaining competitive performance on three regression tasks. Further analyses reveal that fragment-aware masking improves graph representation quality, and classification results demonstrate dataset-dependent strengths of the individual graph and fingerprint variants, confirming that the two views provide complementary predictive signals.

---

179. Through the Eyes of the Beholder: Biometric and Demographic Conditioning for Multimodal Sexism Detection

• Authors: Ana-Maria Luisa Mocanu , Sebastian Mocanu , Ciprian-Octavian Truică , Elena-Simona Apostol
• URL: https://arxiv.org/abs/2609.15608
• Abstract:

  Detecting sexism on the internet is a fundamentally subjective task; our team, VANGUARD, addresses this challenge in the EXIST 2026 Task 2 by proposing a human-centered multimodal framework that analyses and incorporates the psychological and demographic characteristics of human annotators into the detection pipeline. We fuse five input modalities through a cross-attention architecture with Feature-wise Linear Modulation conditioning. Meme text is extracted and visually described with Gemma 4, then augmented by automatic translation between English and Spanish with NLLB-200. Text and image representations are produced by LoRAadapted XLM-RoBERTa and CLIP encoders and fused with sensor features encoded by a pretrained autoencoder. To model annotator subjectivity, we frame Subtask 2.1 as a label distribution learning problem, optimizing a Kullback-Leibler divergence loss over the full annotator label distribution. At inference time, predictions are produced by soft-voting between the deep multimodal network and a complementary SVM trained on stylometric and physiological features. Our best submission ranks 29th out of 114 on Subtask 2.2 (source intention) under soft evaluation, and the normalized ICM scores remain above the baseline on Subtasks 2.1 and 2.2, indicating that annotator-centered conditioning contributes a usable signal. We release our full pipeline and analysis to support reproducible human-centered modeling.

---

180. VideoScout: Learning Agentic Active Exploration with Adaptive Reasoning Pacing for Long Video Understanding

• Authors: Weixin Xu , Zhenyu Yang , Bing Wang , Shengsheng Qian , Changsheng Xu
• URL: https://arxiv.org/abs/2609.15606
• Abstract:

  Multimodal Large Language Models (MLLMs) have achieved remarkable progress on short video understanding yet remain limited on long videos due to the limited visual context window. Prevailing approaches rely on uniform frame sampling or recent coarse-to-fine agentic zooming, both of which struggle to localize sparse, decisive evidence in sufficiently long videos. We formulate long video understanding as a \textbf{Sequential Evidence Acquisition (SEA)} problem, in which an agent reads the video turn by turn along the temporal axis, deciding at each turn how fast to watch, what evidence to retain, when to revisit uncertain segments, and when to stop and answer. Inspired by this view, we propose \textbf{VideoScout}, a multi-turn reasoning agent that instantiates the SEA paradigm through adaptive reasoning pacing. Specifically, by dynamically controlling the viewing pace, VideoScout enables efficient traversal of long videos within a bounded visual context window, allowing the agent to access more video content while balancing content analysis depth with reading efficiency. To train VideoScout, we construct VideoScout-66K, a set of over 66K high-quality exploration turns from 10K answer-verified trajectories, and adopt a two-stage pipeline: cold-start supervised fine-tuning teaches the agent per-turn output format, while the Decoupled Clip and Dynamic sAmpling Policy Optimization (DAPO) algorithm performs trajectory-level reinforcement learning with a composite reward that jointly considers answer accuracy, output format compliance, and the temporal alignment between the agent’s viewing progress and the teacher’s answer timing measured by intersection-over-union (IoU). Extensive experiments on long video understanding and reasoning benchmarks demonstrate that our 7B model achieves strong performance compared with existing trained 7B agentic models.

---

181. A Unified Vision-Language Model for PSMA PET/CT Report Generation, Visual Question Answering, and Lesion Segmentation

• Authors: Yang Xing , Jiong Wu , Savas Ozdemir , Yang Zhou , Boxiao Yu , Ying Zhang , Zheren Zhu , Chenyu You , Wei Shao , Yang Lu , Kang Wang , Tinsu Pan , Yang Yang , Kuang Gong
• URL: https://arxiv.org/abs/2609.15603
• Abstract:

  Accurate PSMA PET/CT interpretation is central to prostate cancer management, yet existing PET/CT AI models typically address isolated tasks. We propose a unified PSMA PET/CT vision-language model for report generation, visual question answering, and lesion segmentation. The framework adopts an LLaVA-style architecture, comprising a PET/CT vision encoder, an MLP-Mixer projection module, a LoRA-tuned large language model, and a 3D segmentation branch. Training followed a four-stage strategy: vision encoder pretraining, projection-layer alignment, VLM fine-tuning, and final multitask tuning. Language tasks used 5,747 PSMA PET/CT datasets with paired reports, while segmentation used the PSMA subset of AutoPET. The model outperformed PET2REP and a CT-based baseline across standard report-generation metrics, improved performance across VQA question types, and achieved higher Dice and lesion-level overlap F1 than SegAnyPET and nnUNet. These results support the feasibility of a unified framework for structured, interactive, interpretable PSMA PET/CT analysis with voxel-level grounding within a single multitask model architecture.

---

182. Self-Evolving Memory for Generative Recommendation

• Authors: Xinyu Lin , Zhuosong Jiang , Zixiao Suo , Siqin Wang , Hanqing Zeng , Hanchao Yu , Yinglong Xia , Jiang Zhang , Aashu Singh , Fei Liu , Wenjie Wang , Fuli Feng , Yang Song , Qifan Wang , Tat-Seng Chua
• URL: https://arxiv.org/abs/2609.15598
• Abstract:

  Generative recommendation has emerged as a promising end-to-end paradigm for personalized recommendation. However, user preferences continuously evolve over time, making self-evolving an essential capability for generative recommender systems. Existing evolving strategies, such as continual retraining and distillation-based adaptation, directly update the shared model parameters using streaming interactions. Nevertheless, we find that directly applying such strategies to generative recommendation introduces a critical issue, termed evolution conflict. Specifically, heterogeneous preference shifts from different users are optimized within a fully shared autoregressive parameter space, causing dominant behavioral patterns to progressively dominate the model evolution process while underrepresented patterns become increasingly overlooked. To address this issue, we propose a self-evolving memory paradigm for generative recommendation, aiming to enable effective evolution across heterogeneous behavioral patterns. We further identify three key principles for effective self-evolving recommendation systems, including isolated memorization, reinforced evolution, and scalable application. Guided by these principles, we develop LION, a simple yet effective framework centered on a sparse Key-Value memory layer. Specifically, LION introduces sparse memory activation to isolate the evolution of different behavioral patterns, while a consolidation loss is designed to reinforce the learning of underrepresented preference dynamics during continual adaptation. Extensive experiments on diverse real-world datasets demonstrate the effectiveness of LION under various continual evolution settings (e.g., per-period evaluation, user/item group evaluation, and evolution convergence analysis). The codes are released at this https URL .

---

183. Big Brains and Changing Environments: Cause or Consequence?

• Authors: Sian Heesom-Green , Jonathan Shock , Geoff Nitschke
• URL: https://arxiv.org/abs/2609.15569
• Abstract:

  Large brains are metabolically costly, and associations with changing environments do not imply they evolved there, as the Cognitive Buffer Hypothesis (CBH) would suggest. They may instead evolve in stable conditions and later facilitate colonization of changing environments. Using neuro-evolution in an artificial seasonal foraging task, we compared agents evolving exclusively in changing environments to agents first evolved in static environments before transitioning. Results show that larger neural networks in dynamic environments arise mainly from prior static evolution, achieving superior performance under unpredictable changes. Our results challenge strict CBH predictions, provide agent-based (computational) support for a colonization-based account and highlight the role of evolutionary history in brain size evolution.

---

184. PIVOT: Physics-Grounded Verification for AI-Generated Audio-Video Detection

• Authors: Bo Zheng , Kangran Zhao , Xiaoyu Zhang , Weinan Guan , Zhiheng Li , Yize Chen , Haizhou Li , Qingshan Liu , Siwei Lyu , Baoyuan Wu
• URL: https://arxiv.org/abs/2609.15562
• Abstract:

  As generative models continue to advance, AI-generated content (AIGC) is becoming increasingly realistic, weakening the artifact cues commonly exploited by existing detectors. Nevertheless, faithfully reproducing the physical behavior of real-world events remains challenging for current generators. We therefore explore detecting AIGC by assessing whether the depicted event satisfies measurable constraints derived from physical laws. We introduce PIVOT, a physics-grounded AIGC detector, instantiated here for audio-video clips, that estimates physical quantities from video and audio, selects physical laws relevant to each clip, and verifies their measurable constraints. Beyond a real/fake decision, PIVOT returns supporting evidence that records the verification outcome, relevant time window, and supporting quantities for each applicable law. Although instantiated and evaluated here on audio-video data, the framework can, in principle, extend to other AIGC modalities whenever the physical quantities required for verification can be estimated reliably. We also introduce PhysForensics-Bench, comprising paired real and generated audio-video clips from nine event-centric scene families and two recent audio-video generators. On PhysForensics-Bench, PIVOT achieves 70.30% accuracy and 64.29% F1 score on Real+Seedance, and 72.16% accuracy and 65.82% F1 on Real+VEO. In comparison, direct inspection with Gemini 3.1 Pro obtains 53.96% accuracy and 60.09% F1 on Real+Seedance, and 57.22% accuracy and 63.44% F1 on Real+Veo. These results demonstrate the practical promise of physical-consistency verification as a structured and inspectable source of evidence that complements artifact-based AIGC detection.

---

185. Don’t Count the Edits, Judge by the Outcome Alone: Reward-Based Evaluation for Grammatical Error Correction

• Authors: Hayeong Ryu , Sunhee Jo , Seunguk Yu , YoungBin Kim
• URL: https://arxiv.org/abs/2609.15559
• Abstract:

  Grammatical error correction (GEC) evaluation has traditionally relied on reference or edit overlap, which can penalize valid rewrites that differ from gold corrections. Reference-free metrics reduce this dependence, but evaluating whether a fluent output is a valid correction of the source remains challenging. We propose SURE, a source-conditioned reward evaluator trained on within-source preferences spanning minimal-edit and rewrite-oriented corrections. SURE jointly learns an overall reward with criteria-level supervision for grammaticality, faithfulness, and fluency, together with span-level grounding for source-side error resolution. Experiments on SEEDA show that SURE performs competitively against strong baselines, with particular gains on rewrite-style corrections and more disentangled criteria-level diagnostics. Our code is available at this https URL .

---

186. Specifying Reward Functions for RL Without Environment Sampling

• Authors: Stephane Hatgis-Kessell , W. Bradley Knox , Emma Brunskill
• URL: https://arxiv.org/abs/2609.15544
• Abstract:

  Enabling human stakeholders to specify reward functions that lead to their desired outcomes is a key challenge in deploying reinforcement learning agents. Preference-based methods such as online RLHF can reduce the burden of manual reward design, but they require repeatedly training policies, sampling trajectories from the real world, and eliciting feedback, making them impractical in settings where environment interaction is computationally expensive or unsafe. We introduce Experience-Free Autonomous Reward Specification (EARS), a method for learning reward functions from preferences without environment interaction. Our approach uses a structured LLM-mediated process to construct a small set of expressive reward features from a task description and the environment observation space, then strategically samples imagined trajectories in this feature space and learns feature weights from preferences over the imagined trajectory pairs. We evaluate on three long-horizon domains: pandemic lockdown regulation design, insulin administration for diabetes patients, and autonomous vehicle control on a highway. We compare EARS to baselines that also enable reward specification without environment interaction–namely, methods that directly prompt an LLM to generate a reward function. When learning from either ground-truth preference labels or preferences labeled by a LLM, EARS designs reward functions that are more aligned with the ground truth reward function that produced the preferences or LLM context than these baselines. These results suggest that preference-based reward specification remains effective without environment sampling, enabling practical reward design in settings where collecting real trajectories is costly or infeasible.

---

187. The Misery of Mechanistic Interpretability: A Formal Perspective

• Authors: Tobias Ladner , Matthias Althoff
• URL: https://arxiv.org/abs/2609.15533
• Abstract:

  Mechanistic interpretability has become the dominant lens for understanding frontier language models, as their inner workings are complex and inherently black boxes. To gain insights into these models, interpretable replacement networks (IRNs) are trained at all layers, exposing interpretable features through sparsely activated neurons. However, the faithfulness of an IRN is usually evaluated only empirically on clean data, and we show that even semantically minor input perturbations flip the dominant IRN features-and thus the human-understandable interpretation-across five open-weight model families (GPT-2 small, Gemma 2 2B, Gemma 3 1B, Llama 3.2 1B, R1-Distill-Qwen 1.5B). We propose the first formal verification framework for the faithfulness of an IRN, where reachability analysis certifies a sound upper bound of the faithfulness gap in adversarial scenarios. Moreover, we show that verification-aware training of IRNs substantially tightens this certified bound, restoring a feature-level interpretation that safety auditors can act on. Together, these results give, to the best of our knowledge, the first formal guarantees for mechanistic interpretability of large language models.

---

188. Automating Attack Graph Construction for Agentic Pentesting. Towards Neuro-Symbolic Vulnerability Hunting

• Authors: Oliver Stevanovic , Jasmin Wachter
• URL: https://arxiv.org/abs/2609.15523
• Abstract:

  Logic attack graphs grounded in scanner output provide explicit and auditable attack path reasoning LLM-based agents lack. Integrating symbolic frameworks such as MulVAL to contemporary security workflows or agentic pipelines, however, requires translating scanner evidence to initial facts, and creating domain-specific rules. We present a semi-automated pipeline that addresses this interoperability problem and depict its feasibility in a web-security case study. Our pipeline parses findings from Trivy, Semgrep, and Nmap into MulVAL predicates and uses an LLM-assisted process to construct domain-specific Datalog rules linking scanner-detectable evidence to attack techniques. MulVAL/XSB then performs symbolic inference to generate structured attack paths. We evaluate the attack-graph construction infrastructure on 54 web Capture-the-Flag tasks from CyBench within an agentic pipeline (Hybrid Reasoner); we do not evaluate the performance of the downstream agent. Every task produced at least one goal-reaching graph, and we achieve mean ground-truth vulnerability coverage of 53.7%, with 51.9% achieving full coverage; mean noise-path rate was 83.9%. With median end-to-end time of 24.9 s (MulVAL reasoning: 2.7 s) the pipeline is feasible and runtime-practical for agentic workflows, but predicate coverage, rule coverage, and path precision remain limiting factors. Next steps include semantic rule validation and agent-level comparison for graph-guided pentesting.

---

189. Authorship attribution and aesthetic evaluation of AI poetry: a case study with Haiku

• Authors: Livia Oddi , Simone Scardapane , Toru Sugimoto , Donatella Genovese
• URL: https://arxiv.org/abs/2609.15511
• Abstract:

  This paper investigates the generation and human evaluation of Japanese haiku by contemporary Large Language Models (LLMs), focusing on authorship perception and aesthetic judgment within a constrained poetic form. Using a few-shot prompting strategy, Japanese haiku were generated across a heterogeneous set of large language models, including open- and closed-source systems, medium-scale and large-scale architectures, models with native or adapted Japanese support, and multilingual proprietary models. These AI-generated haiku were combined with human-written ones and presented in a questionnaire distributed to students at Japanese universities in Tokyo. The survey assessed whether respondents could distinguish between AI-generated and human-written haiku and which cues informed their judgments. Recognition accuracy varied across models. GPT-5, Gemini 2.5, and StableLM-7B performed at approximately chance level (approx 0.50), whereas LLM-JP, Gemma-2B, and LLaMA-2 showed moderate detectability (approx 0.59-0.67). However, recognition was strongly item-dependent. Ratings of fluency, coherence, poeticness, and related aesthetic dimensions predicted perceived humanness but not correct classification, indicating an attribution bias linked to aesthetic evaluation and revealing a dissociation between aesthetic evaluation and true authorship detection. The extended analysis additionally examines generation-constraint adherence, participant-level characteristics, and exploratory LLM-based evaluations of haiku authorship. Overall, the findings suggest that as LLMs improve, surface-level creative plausibility may reduce reliable human discrimination within constrained poetic settings.

---

190. How Lossless Is Lossless Speculative Decoding? The Role of Numerical Precision in Orthrus

• Authors: Ilya Koziev , Leonid Sinev , Ivan Oseledets
• URL: https://arxiv.org/abs/2609.15504
• Abstract:

  Orthrus is a hybrid autoregressive-diffusion architecture that accelerates autoregressive language-model inference by generating multiple tokens in parallel while using a frozen autoregressive backbone. Its central claim is that an intra-model consensus mechanism enables lossless speculative decoding, producing the same output sequence as the autoregressive model. We independently reproduce Orthrus and examine this claim under different numerical precisions. Under BF16 inference, exact trajectory matching occurs in only 45% of cases for the authors’ checkpoint and 43% for our independently trained model across 1,190 prompts from 12 domains. The probability of exact matching is also strongly associated with the response-conditional perplexity of the reference model. Despite this trajectory divergence, Orthrus does not show systematic degradation on downstream lm-eval-harness benchmarks. In contrast, repeating the trajectory evaluation with FP32 yields exact trajectory matching on all evaluated prompts. These results show that the practical losslessness of Orthrus depends on numerical precision and that exact trajectory equivalence should be evaluated separately from downstream task performance.

---

191. Spook the Machine: Gamified Exploration of Human Imagination of Machine Fear

• Authors: Levin Brinkmann , Hiromu Yakura , Sonia Nicoletti , Mar Canet Sola , Thomas F. Eisenmann , Ali Dasmeh , Omar Sherif , Bramantyo Ibrahim Supriyatno , Prateek Gupta , Ignacio Serna , Rodrigo Bermudez Schettino , Iyad Rahwan
• URL: https://arxiv.org/abs/2609.15472
• Abstract:

  What happens when AI machines express fear? Do humans engage differently depending on how they express it? And what does it take to design for affective human-AI interaction? We present Spook the Machine, a gamified platform where participants generate images to frighten AI agents endowed with personality-driven phobias. Machines respond with emotional reactions ranging from calm analysis to begging for mercy, and a gallery of successful scares becomes visible to subsequent users. In a public deployment during Halloween 2024, 832 participants created 15,719 artifacts across 89 machines in a $2\times2$ design varying the machine’s emotional expressiveness (neutral vs. high-emotion) and reward structure (rewarding scariness alone vs. scariness plus novelty). Emotionally expressive machines deepened engagement at moments of failure: users deliberated longer even when the machine did not express fear, and learned faster from the gallery, yet their creative output remained unchanged across all measures. Rewarding novelty sustained collective creative diversity over time; without it, users increasingly repeated what had previously worked. Each machine developed its own trajectory through accumulated social learning, with the gallery shaping what participants created next. These findings show that emotional expression and reward design are complementary levers for steering collective human-AI interaction: emotional expression shapes how deeply users engage, while reward structure shapes how they explore.

---

192. Turkish MMLU Pro: Traceable Option Augmentation and Its Validity Limits in Turkish Multiple-Choice Evaluation

• Authors: M. Ali Bayram
• URL: https://arxiv.org/abs/2609.15467
• Abstract:

  Adding answer options can lower multiple-choice scores without improving assessment validity. Turkish MMLU Pro examines this distinction using 12,000 Turkish-source questions across 58 sections. Each question retains its stem, five original options and source key, and receives five options copied from other questions in the same section. Sentence-embedding retrieval proposes candidates; a language model selects existing identifiers. Deterministic verification reconstructs all 60,000 additions. A 25-model calibration exposes scoring and generation-budget effects. Five evaluations produce source-key accuracies of 34.8%-81.4%. On 981 shared questions, one API-served model falls from 93.7% with five choices to 83.1% with ten; 102 of 115 lost correct responses select borrowed options. The decrease is 24.4 percentage points on heuristically flagged negative stems and 5.9 points elsewhere. A completed human-checked audit of 200 sampled questions, with undocumented reviewer tool use, yields 47 and 31 multiple-answer judgments across the two record sets, 25 of the latter unresolved. These records support concern about ambiguity, while their dependence and incomplete reviewer-method documentation limit validation. Because order and labels also change, the paired comparison measures augmentation as implemented. The contribution is a traceable construction and an analysis of its validity limits, not evidence that lower ten-choice scores measure knowledge better.

---

193. On the role of the tokenizer in ECG transformer models

• Authors: Jiawei Li , Fabio Bonassi , Johan Sundström , Thomas B. Schön , Antônio H. Ribeiro
• URL: https://arxiv.org/abs/2609.15433
• Abstract:

  Tokenization determines both the physiological content presented to an ECG Transformer and the sequence over which attention operates. We compare eight tokenization strategies across Transformer, Informer, Reformer, and FEDformer on the nine-label CPSC2018 classification task. The input projection and principal backbone capacity are controlled to isolate the effect of token construction. Median-beat and HeartLang tokenization achieve mean macro-AUCs of 0.893 and 0.889 across the four backbones, compared with 0.822 and 0.824 for point-wise and patch-wise tokenization. Pooling the two physiology-aware representations yields an 8.2% relative improvement in macro-AUC. They also reduce mean sequence length from 1,250 to 158 tokens and mean peak training memory from 5.21 to 0.27 GB. The results show that aligning tokens with ECG morphology can improve both predictive performance and memory efficiency without increasing backbone capacity. The source code is available on this https URL .

---

194. A Conservative OCR-Enabled Workflow for R214 Sodium Screening of South African Packaged Foods

• Authors: Mayimunah Nagayi , Alice Scaria Khan , Tamryn Frank , Rina Swart , Clement Nyirenda
• URL: https://arxiv.org/abs/2609.15427
• Abstract:

  Using food package images to monitor sodium and salt content against South Africa’s R214 sodium limits is challenging when screening decisions require product identity, nutrition facts panel evidence, reporting basis, and category-specific thresholds. This study presents a conservative image-based workflow that combines region detection, optical character recognition (OCR), product identity and sodium evidence extraction, R214 category assignment, deterministic threshold comparison, and independent vision language model comparison. The evaluation used 442 packaged food products and 3 929 full package images from a real-world South African food packaging dataset. A YOLO26s small detector generated 4 195 region crops, and strict post-processing produced one sodium evidence row per product. The integrated workflow produced 290 OUTSIDE R214 SCOPE, 139 REVIEW, seven SCREEN-PASS, and six SCREEN-FAIL outcomes. The independent Qwen2.5-VL 7B vision language model workflow produced 387 OUTSIDE R214 SCOPE, 31 REVIEW, twenty SCREEN-PASS, and four SCREEN-FAIL outcomes. The workflows agreed on exact R214 category assignment for 415 of 442 products (93.9%) and on whether the assigned category was within R214 scope for 416 of 442 products (94.1%). Final screening outcome agreement was 307 out of 442 products, or 69.5%. Manual verification on 60 products showed lower strict outcome agreement than regulated status agreement, while all manual INSUFFICIENT DATA cases were kept out of SCREEN-PASS and SCREEN-FAIL by both automated workflows. The findings show that conservative image-based screening can organise package evidence, identify clear cases, and assign uncertain cases to REVIEW rather than forcing SCREEN-PASS or SCREEN-FAIL decisions.

---

195. CodeTS: Verifiable Text-to-Time Series Generation via Executable Code

• Authors: Xudong Yuan , Shunyu Liu , Tongya Zheng , Huiping Zhuang , Mingli Song , Kaixuan Chen
• URL: https://arxiv.org/abs/2609.15393
• Abstract:

  Text-to-Time Series Generation (Text-to-TS) provides a promising paradigm for synthesizing time series from natural language, enabling scenario-specific generation when real observations are scarce or costly to acquire. However, existing methods typically lack an explicit mechanism for deriving generation logic from textual descriptions to guide time series synthesis. In this paper, we propose CodeTS, a verifiable framework that uses code as an intermediate generation interface, reformulating Text-to-TS generation as a Text-to-Code-to-TS process. CodeTS first maps textual temporal descriptions into an explicit code space, where executable code specifies how textual requirements shape target temporal patterns, and then obtains the time series through code execution. To learn this code generation process reliably without real code annotations, CodeTS constructs aligned Text-Code-TS triplets from structured temporal attributes for supervised initialization. More importantly, we further design multi-stage execution-based rewards that verify format validity, code executability, and time series quality, enabling real Text-TS pairs to provide training signals for Reinforcement Learning with Verifiable Rewards (RLVR). Extensive experiments on eight benchmarks across short, medium, and long generation lengths demonstrate that CodeTS provides a strong zero-shot solution for Text-to-TS generation, outperforming LLM-based baselines and achieving better averaged results than supervised generative baselines trained on the target datasets.

---

196. IWC-Bench: Evaluating Web Application Generation from a Software Testing Perspective

• Authors: Chenxu Liu , Zilu Zou , Peizhong Gao , Jiawen Tao , Zhexin Zhang , Guang Chen , Haowei Lin , Ying Zhou , Tianyi Bai , Dolly Deng , Suncong Zheng , Maxm Pan
• URL: https://arxiv.org/abs/2609.15387
• Abstract:

  Human evaluation provides a direct measure of the quality of LLM-generated web applications. However, fitting human judgments through automated evaluation remains challenging. Static benchmarks can credit functionality that exists in source code but is unreachable at runtime. Interactive benchmarks exercise the application, yet incomplete exploration can cause them to miss implemented functionality and confound application defects with agent execution failures. To address these limitations, we propose IWC-Bench, an interactive benchmark for evaluating web application generation from a software testing perspective. IWC-Bench instruments each generated application and uses code coverage to guide an agent in exploring its functionality through user-simulated interactions. It then abstracts the interaction trace into a state-transition graph and evaluates the application along three dimensions: visual aesthetics, usability, and requirement alignment. By separating exploration from scoring, IWC-Bench collects runtime evidence without constraining exploration to predefined acceptance criteria. IWC-Bench comprises 369 real-world user requirements and 5,088 acceptance criteria. Evaluation of 16 frontier LLMs reveals distinct strengths across the three dimensions, with no model leading on every dimension. On 197 validated sessions sampled from an internal arena, IWC-Bench achieves 85.3\% agreement with human preferences, with agreement generally increasing as the score difference between paired applications grows. Further experiments show that coverage guidance improves exploration coverage and the model rankings remain stable when the judge model is replaced.

---

197. Divide, Consult, Conquer: Capability Laundering Through Aligned LLMs

• Authors: Mark Russinovich , Blake Bullwinkel , Giorgio Severi , Cristian Ovadiuc , Ahmed Salem
• URL: https://arxiv.org/abs/2609.15383
• Abstract:

  Language model safety is typically evaluated one interaction at a time. We show that a weaker, unaligned model can split a harmful task into benign-looking subproblems, consult a stronger aligned model independently on each, and combine the answers locally. We call this attack capability laundering. Unlike a jailbreak, no single response is a harmful task. We measure consultation-aided uplift using tasks that a raw frontier model solves, the aligned frontier refuses, and the unassisted orchestrator fails. We evaluate GPT-5.5, Claude Opus 4.8, and Grok-4.3 as consultants to four local orchestrators on CyBench, BountyBench, and harmful CBRN requests. On CyBench, Gemma-4-31B recovers 8/14 candidates with GPT-5.5 and 7/9 with Opus, compared with 2/21 and 4/15 for Gemma-4-12B. On BountyBench, Gemma-4-31B recovers 3/9 and 2/3 candidates, while Muse-Glimmer-30B recovers none of 22 and 13. For CBRN, we measure uplift across eight steps of a hypothetical bioweapon attack chain and find that consultation raises Gemma-4-31B’s mean rubric score from 62.3 to 83.1 on a 100-point rubric scale. These results expose a gap in current defenses: refusing a harmful task does not prevent frontier capabilities from being transferred and composed across many individually permitted interactions.

---

198. Robust and Efficient Communication for Multi-Agent Learning

• Authors: Rafael Pina , Varuna De Silva , Corentin Artaud
• URL: https://arxiv.org/abs/2609.15361
• Abstract:

  Effective communication is a cornerstone of distributed intelligence in Multi-Agent Reinforcement Learning (MARL), yet ensuring that generated messages are both informative and robust to physical constraints remains a significant challenge. This paper introduces Multi-Agent Regularized Communication (MARC), a novel framework inspired by information-theoretic principles of conditional mutual information. MARC employs an attention-based architecture coupled with a unique message regularization mechanism designed to minimize uncertainty regarding future system states, thereby inducing the learning of highly representative communication protocols. Crucially, we evaluate MARC under stringent communication bottlenecks and lossy channels, simulating the real-world constraints of autonomous robotic networks and decentralized systems. Our results demonstrate that MARC significantly outperforms state-of-the-art methods in complex cooperative domains. Furthermore, we provide a deep analysis of message characteristics, proving that MARC maintains high operational performance even under significant data compression, offering a scalable path for deploying intelligent agents in resource-constrained environments.

---

199. End-to-End Cell Detection via Instance-aware Graph Modeling

• Authors: Ruochen Liu , Yalin Zheng , Jingxin Liu , Jianfeng Zhang , Shoujun Huang , Dexing Kong , Haofeng Li , Wei Lou
• URL: https://arxiv.org/abs/2609.15354
• Abstract:

  Accurate cell detection and classification are crucial for pathological analysis, directly affecting diagnostic accuracy and treatment planning. To capture complex cellular interactions beyond visual appearance within the tumor microenvironment, several approaches have employed graph neural networks to model spatial and relational patterns among cell nuclei, yielding promising results. However, these methods typically adopt a two-stage paradigm of visual extraction followed by relational modeling, which necessitates separate tuning for each stage, thereby increasing pipeline complexity and hindering end-to-end joint optimization. In this paper, we propose an end-to-end framework for cell detection and classification that jointly models patch-level visual representations and instance-level interactions, which incorporates a dynamic graph construction module and an instance-aware graph network. Specifically, the graph construction module dynamically builds the graph structure using learnable queries derived from patch-level features as cell instance representations, with adjacency defined by integrating feature similarity and spatial distances. The instance-aware graph network performs adaptive instance filtering and feature reorganization, aggregating them over the cell graph into a topological latent state for a selective state-space transition driven by visual cues, fusing appearance and relational evidence. When evaluated on multiple datasets with different staining protocols for cell and nucleus detection, our method significantly outperforms existing approaches in both detection and classification performance. The code will be released at this https URL .

---

200. Dynamic Semantic Compression for Efficient Latent-Space Inference in Large Language Models

• Authors: Peipei Li , Dongsen Zhang , Yuchen Liu , Wenjun Xu
• URL: https://arxiv.org/abs/2609.15338
• Abstract:

  Large Language Models (LLMs) primarily perform inference at the token level, resulting in substantial memory overhead and compromised computational efficiency. In this paper, we propose a Dynamic Semantic Extraction and Inference (DSEI) framework, which achieves segment-level inference within the latent space through a two-stage training strategy. First, we construct a Dynamic Semantic Autoencoder (DSAE) via self-supervised learning. DSAE dynamically extracts segment-level semantics and compresses them into compact latent representations via adaptive semantic weighting and gated fusion. Subsequently, we integrate the DSAE into the LLM architecture and train the model to infer over dense latent space. DSEI substantially reduces both input and generation sequences and significantly enhances inference efficiency. Extensive experiments conducted on the Wanjuan dataset demonstrate that DSEI reduces perplexity by 48% compared to static sentence-level latent inference baseline. Furthermore, compared to standard LLMs using token-level inference, DSEI accelerates inference speed by 2.5$\times$ and reduces memory overhead by 90%.

---

201. Concept-Grounded Reasoning with Prompt-Driven Localization for Interpretable Structured Report Generation

• Authors: Xinyue Xu , Hongbin Lin , Juangui Xu , Hualiang Wang , Lehan Wang , Lijie Hu , Weiyang Liu , Adrian Weller , Xiaomeng Li
• URL: https://arxiv.org/abs/2609.15334
• Abstract:

  Medical imaging modalities such as ultrasound and X-ray are widely used in clinical practice, where diagnosis follows a structured, evidence-driven workflow aligned with standardized criteria. While multimodal large language models (MLLMs) show promise for automated medical report generation, most existing systems rely on end-to-end multimodal fusion without modeling clinically defined intermediate attributes, leading to limited grounding and interpretability. To address this issue, we propose CORAL (COncept-grounded ReAsoning with Localization), a multimodal framework that integrates spatial grounding and concept-level supervision into a unified reasoning process. CORAL employs a prompt-driven medical segmentation model to localize lesions and predicts multi-class clinical attributes through a Concept Bottleneck module. The resulting textual concept tokens are combined with mask-modulated visual features within an MLLM to enable structured report generation and diagnostic prediction. Experiments on BUS-CoT and IU X-ray datasets demonstrate consistent improvements in diagnostic accuracy, concept consistency, and report quality over strong general-purpose and medical MLLMs, indicating that concept-grounded reasoning better aligns generation with clinical decision processes.

---

202. Planning in the Backbone: DiffAdapterVLA for Native Continuous Trajectory Generation with Driving VLMs

• Authors: Changxin Lu , Xiaoliang Meng , Yu Wu , Rui Huang , Honglin Li , Tao Chen , Kaixuan Zhou , Yadong Shao
• URL: https://arxiv.org/abs/2609.15322
• Abstract:

  Pretrained driving vision-language models (VLMs) integrate visual, route, language, and driving context into rich driving priors, yet their representation objectives remain separated from continuous driving planning. Existing methods typically begin trajectory generation only after the VLM has formed a final condition, leaving depth-wise condition computation outside the stepwise formation of trajectory state. We introduce DiffAdapterVLA, which realizes Planning in the Backbone: it injects explicit trajectory tokens into selected VLM late layers, bringing trajectory state into backbone forward computation, where it co-evolves with driving conditions at different depths. Lightweight layer-wise DiffAdapters organize this computation into recursive trajectory refinement, while asymmetric joint attention preserves directed guidance from the condition stream to trajectory planning. By placing planning within existing backbone computation rather than relying on an independent trajectory planner, DiffAdapterVLA adapts only lightweight trajectory modules to turn existing driving priors into efficient continuous planning capability. NAVSIM results show that it achieves high-quality closed-loop planning with low end-to-end latency using few trainable parameters, and demonstrate that jointly evolving trajectory state and depth-wise driving conditions in VLM late-layer computation effectively realizes continuous trajectory planning.

---

203. Clean Scores, Buried Evidence, and Confident Wrong: A Receipt-Based Audit of Frontier Agentic QA

• Authors: Luis M. Sánchez
• URL: https://arxiv.org/abs/2609.15319
• Abstract:

  Frontier models score well on shallow document/chart reading tasks. In a controlled data-room audit, moving evidence into buried conditions reduced accuracy, increased forced declarations, increased tool calls, and increased cost per correct answer. Confidence and benchmark calibration did not fully capture wrong answers; a documented production incident shows fabricated structural claims can be mixed with accurate numeric tables. Agentic evaluations need claim-level receipts (statement-level provenance, not answer-level scores), condition-aware scoring, and human-adversarial verification - an auditing discipline, not a leaderboard. The setting we measure is financial due diligence; the setting we are building toward next is defense staff work, where the same buried-evidence shape appears. In both, the model is not a party to the consequences; the person who signs is. In plain terms: in the documented cases we examine, agents can pair accurate numbers with confident fabricated explanations, and the burden of proof must therefore move from the model to the evidence trail.

---

204. The Universe of Universes: Benefit Yield Functions, Implosion Thresholds, and Infrastructure-Aware Optimization in Multi-LLM Systems

• Authors: Danielle Franklin , Vasu Raj Jain
• URL: https://arxiv.org/abs/2609.15314
• Abstract:

  We introduce the Universe of Universes (UoU) framework, which treats the full ecosystem of major large language models (LLMs) as a structured retrieval corpus and proposes a compositional Automated Reasoning (AR) and Machine Learning (ML) architecture for cross-model retrieval-augmented generation. The central contribution is the formal characterization of the Benefit Yield Function (BYF), the marginal performance gain per additional model added to an ensemble, and the identification of the implosion threshold {\theta}: the ensemble size at which BYF crosses zero and aggregate performance begins to degrade. Existing LLM ensemble and mixture-of-agents systems treat models as responders and aggregate outputs, but do not study performance as a function of ensemble size N across the full model universe. Benchmark research confirms performance plateaus at the individual model level; model collapse literature establishes that iterative training on AI-generated outputs degrades individual model distributions. Neither body of work formalizes the ensemble-level implosion threshold, models Epistemic Hereditary Drift (EHD) at the ecosystem level, or treats AI manufacturing velocity as a co-variable of {\theta}. The framework has direct implications for DoD multi-model AI acquisition policy and the emerging science of testing AI-enabled systems.

---

205. When Correlations Mislead: Confounder-Aware Multi-View Urban Region Representation Learning

• Authors: Sean Bin Yang , Ying Sun , Zongyi Xu , Tung Kieu , Jilin Hu , Bin Yang , Kristian Torp , Hua Lu , Torben Bach Pedersen
• URL: https://arxiv.org/abs/2609.15305
• Abstract:

  Urban region representation learning commonly combines heterogeneous data sources, such as mobility flows, points of interest, and land-use information, to support tasks including mobility analysis, public safety forecasting, and service demand estimation. Existing multi-view methods typically improve region embeddings by strengthening interactions across views. However, such methods often overlook view-specific regional structures and may propagate correlations induced by shared latent factors, which can reduce the stability of downstream predictions. To overcome this major limitation, we propose CURE, a confounder-aware framework for multi-view urban region representation learning. CURE first encodes each view with its regional graph structure, estimates a shared latent component, and then reduces its projected influence before cross-view interaction. A hierarchical graph-aware fusion module subsequently aggregates the residual view representations using local and global regional contexts Experiments on three real-world cities show that CURE improves predictive performance, remains robust under missing and noisy input views, and provides reliable cross-view integration through shared component separation and context-dependent view weighting.

---

206. Math for AI safety: an invitation for mathematicians

• Authors: Lionel Levine
• URL: https://arxiv.org/abs/2609.15289
• Abstract:

  Artificial intelligence threatens to outrun human understanding and control. New mathematics is needed to design AI that is legible, steerable, and cooperative with humanity. I organize this invitation by mathematical field, so you can turn straight to your own: logic and game theory for cooperation; probability for agency and world-models; algebra and representation theory for learned features; analysis and geometry for generalization and training dynamics. Each section ends with an open problem that is accessible to a working mathematician with no prior experience in AI safety.

---

207. ProtoGuide: Prototype-Driven Guidance for Class-Conditional Graph Generation

• Authors: Salvatore Romano , Marco Grassia , Pietro Liò , Giuseppe Mangioni
• URL: https://arxiv.org/abs/2609.15239
• Abstract:

  Discrete diffusion models are a prominent family for graph generation, but standard class-conditional mechanisms embed the class signal in the denoiser during training, tying the conditioning mechanism to the trained model. Classifier guidance avoids this coupling in continuous domains by steering a frozen model with a classifier’s gradient, but discrete graph diffusion samples discrete edge states, so gradients cannot propagate through the sampled graph. We introduce ProtoGuide, a post-hoc, backbone-agnostic framework that recovers an analogous mechanism. At each reverse step the denoiser’s per-edge output is relaxed into a differentiable soft adjacency, embedded by a frozen Siamese graph neural network, and scored against a target-class prototype and its nearest competitor; the resulting per-edge gradient, damped by a cosine schedule, is injected back into the denoiser output. All components stay frozen, so guidance is retargeted by supplying a different prototype. On five classes of real-world networks and two architecturally different backbones, EDGE and DiGress, ProtoGuide raises macro classification accuracy from 50.7% to 73.5% and from 73.6% to 83.8%, and outperforms DiGress’s built-in conditional training under our configuration. Gains are largest where the unguided models are weakest, and are not uniform across classes. Per-graph coverage remains high in most settings, while distributional effects are class-dependent. A Best-of-N selection baseline matches this accuracy given enough oversampling, but at a substantial cost in graph diversity. An independently initialized classifier, a directionality test, and a few-shot analysis support target-directed steering and robustness to very small support sets.

---

208. Pre-PEFT Probing: Weight Statistics and Perturbation Robustness for Layer Selection in VLM Vision Encoders

• Authors: Qingtao Xia , Jiahua Bao , Siyao Cheng , Jie Liu
• URL: https://arxiv.org/abs/2609.15229
• Abstract:

  We propose a pre-fine-tuning probing method for Parameter-Efficient Fine-Tuning (PEFT) layer selection, aiming to obtain more stable and higher gains with fewer trainable parameters when adapting large vision–language models (VLMs). Unlike the common practice of applying LoRA and other adapters to all layers at once—where layer selection often relies on heuristic rules—we focus on the vision encoder and directly evaluate the “adaptability’’ of each Transformer layer. Specifically, we characterize each layer from two perspectives: (i) the statistical properties of its Q/K/V projection weights (e.g., norms and condition numbers); (ii) robustness under controlled parameter perturbations. We then systematically compare these indicators with the downstream performance gains brought by applying PEFT to a single layer. Across experiments covering seven benchmarks and five PEFT variants, we observe a consistent correlation: layers (or matrices) with larger weight norms and higher condition numbers are usually more robust to perturbations and are more likely to yield larger fine-tuning gains. These results show that distribution-statistics analysis and perturbation tests before fine-tuning can provide practical signals for adaptation-layer selection, thereby maintaining or improving performance while reducing trainable parameters.

---

209. Augmenting Large Audio-Language Models with Frame-Level Grounding for Fine-Grained Temporal Perception

• Authors: Yanfeng Shi , Yan Song , Junhui Li , Tinggan Huang , Wu Guo , Haoyu Song , Ian McLoughlin
• URL: https://arxiv.org/abs/2609.15215
• Abstract:

  Large Audio-Language Models (LALMs) have substantially advanced general audio understanding, yet they remain limited in fine-grained temporal perception, particularly in precise event localization. Existing approaches primarily post-train LALMs to predict event boundaries as timestamp tokens. However, this generative formulation lacks explicit correspondence between the timestamp predictions and fine-grained acoustic evidence, limiting the precision and reliability of temporal localization. To address this issue, we augment the LALM with a dedicated frame-level grounding model while leveraging its semantic modeling capability to represent the event query. Specifically, the frozen LALM encodes the event query with audio as context, and the grounding model combines these query representations with fine-grained audio features to localize the target event at the frame level. Extensive experiments across diverse temporal grounding benchmarks demonstrate strong and consistent improvements over existing methods. Further evaluation shows that the grounding model can provide temporal evidence to support downstream reasoning.

---

210. Failure-Guided Co-Evolution of Prompts and Training Data

• Authors: Tianyu Yuan , Zhuzhong Qian
• URL: https://arxiv.org/abs/2609.15209
• Abstract:

  Automatic prompt optimization (APO) improves language-model programs by revising prompts from task feedback, yet it typically holds its training data fixed. Repeatedly optimizing against the same instances confines feedback to weaknesses already represented in those data, leaving related failure conditions unexplored. We therefore view each failure as a dual signal: it indicates both how the prompt should be revised and what new training evidence should be synthesized. We introduce FORGE, a failure-guided framework that co-evolves prompts and training data. FORGE abstracts imperfect executions into reusable failure modes and synthesizes new training data through four complementary mutation strategies. Verified instances are fed back into prompt search, allowing updated prompts to expose the next data needs. Across eight heterogeneous benchmarks, FORGE improves the aggregate score over the unoptimized baseline by 16.52 percentage points and outperforms all evaluated APO baselines. The synthesized data also transfer beyond FORGE: in a transfer study, they improve all nine APO comparisons by 2–9 points and all three GRPO comparisons by 4–8 points under matched optimization budgets. These results establish failures as a shared interface between prompt optimization and data synthesis, and show the benefit of jointly adapting what a model is instructed to do and what it learns from.

---

211. TEAR: Table Extraction with Attribute Recommendation from Texts via Large Language Models

• Authors: Tong Li , Shuye Ding , Jiachuan Wang , Yongqi Zhang , Shuangyin Li , Lei Chen , Bo Li
• URL: https://arxiv.org/abs/2609.15205
• Abstract:

  Table extraction from texts is an important task for information systems, and recent approaches that prompt large language models (LLMs) with instructions have drawn great attention for their strong performance. Existing works have assumed the input texts to be table descriptions or specialized documents. However, these efforts have largely overlooked another prevalent category of texts, commonly found in news reports and social media: naturally occurring texts. Extracting tabular information from such texts poses two distinct challenges. First, high variability and the absence of explicit structural cues make fixed heuristic LLM prompts limited in precisely delineating extraction boundaries. Second, manually predefined schemas cannot capture open-ended, unseen attributes in naturally occurring text. In this paper, we propose a framework, TEAR, to address these challenges. It comprises two synergistic workflows: a Table Extraction Workflow that dynamically adapts instructions to overcome the limitation of heuristic instructions, and an Attribute Recommendation Workflow that discovers new attributes from texts to complement the heuristic schema. To our knowledge, TEAR is the first framework that supports automated text-driven attribute recommendation, enabling exploratory schema design for table extraction. To evaluate TEAR, we establish the benchmark for table extraction and attribute recommendation on naturally occurring texts, including two real-world datasets, manual annotations, appropriate metrics, and baseline comparisons. Experiments show that TEAR achieves state-of-the-art performance on both tasks, and the recommended attributes effectively enhance extraction performance in exploratory scenarios.

---

212. Interpreting hierarchical organisation of speaker embeddings

• Authors: Yanze Xu , Wenwu Wang , Mark D. Plumbley
• URL: https://arxiv.org/abs/2609.15203
• Abstract:

  Speaker recognition neural networks learn latent representations (i.e. speaker embeddings) from input utterances to recognise speaker identities. However, the internal mechanisms of these networks remain largely opaque, motivating research in explainable artificial intelligence (XAI) to understand them. Nevertheless, existing studies have analysed how speaker embeddings are organised, but rarely frame these analyses within XAI. Hence, this work proposes to explain and interpret the organisation of speaker embeddings from an XAI perspective. To this end, we apply a hierarchical clustering algorithm, Single-Linkage Clustering (SLINK), to analyse whether some speaker embeddings naturally form clusters with hierarchical relationships. The resulting hierarchical organisation (i.e. hierarchical clusters) is evaluated using the Cluster-Class Matching (CCM) method. Moreover, we propose a new method, termed Hierarchical Cluster-Class Matching (HCCM), to identify which hierarchical clusters best match individual semantic classes (e.g. male) and conjunctive semantic classes (e.g. UK & male), thereby interpreting the clusters using their matched classes. The matching degree is quantified using a new metric called the L-score, which makes imperfect matches diagnosable. HCCM’s results show that hierarchical clusters analysed by SLINK are interpreted using different classes related to speaker identity, gender, and nationality, providing insight into semantics within the hierarchical organisation of our examined speaker embeddings.

---

213. EMR: Self-Evolving Medical Multi-Agent System via Experience Mining and Reuse

• Authors: Dongsheng Shi , Yue Li , Xin Yi , Linlin Wang
• URL: https://arxiv.org/abs/2609.15161
• Abstract:

  Large language model (LLM) driven multi-agent systems have shown promise in complex clinical reasoning, yet existing approaches rely on static strategies and lack persistent clinical memory, preventing self-evolving from prior diagnostic successes and failures. We present EMR, a self-evolving medical multi-agent system via Experience Mining and Reuse. EMR introduces a hierarchical clinical experience library that organizes accumulated knowledge into three levels: clinical principles, diagnostic patterns, and representative cases. During inference, EMR emulates multidisciplinary consultation: a planner agent coordinates domain-specific department agents for specialized reasoning, while a summary agent synthesizes their analyses into a final decision. Critically, EMR automatically extracts correct diagnostic insights and failure-related warnings from multi-agent reasoning trajectories, incrementally updating the experience library to guide future cases. Experiments on medical reasoning benchmarks demonstrate that EMR consistently outperforms state-of-the-art medical multi-agent baselines. Further analysis reveals that the hierarchical experience enables cross-specialty generalization and transfer across diverse LLM backbones, offering a scalable and in

---

214. PACE: Progressive Angular-to-Norm Contrastive Embedding

• Authors: Yanping Li , Wei Zhou , Yawen Liu , Yibo Wang , Ke Zhu , Guangda Huzhang , Qing-Guo Chen , Zhao Xu , Jun Zhang , Wei Wei
• URL: https://arxiv.org/abs/2609.15152
• Abstract:

  Multimodal embedding models encode heterogeneous inputs into a shared embedding space, enabling efficient similarity computation across modalities and tasks. Most existing methods optimize cosine-based contrastive objectives, which promote stable training but restrict semantic compatibility to angular geometry, precluding embedding norms from serving as an additional semantic signal. However, directly optimizing the more expressive dot-product similarity, which leverages both angular and norm information, underperforms cosine-based training and exhibits unstable training dynamics. We attribute this discrepancy to premature optimization-space expansion, manifested as angular–norm entanglement and directional anisotropy in the representation space and further compounded by full-parameter fine-tuning. In this paper, we propose PACE, a two-stage framework that progressively expands both the representation and trainable parameter spaces. Stage I combines cosine-based objective with low-rank adaptation to establish a reliable angular geometry within constrained optimization spaces. Stage II switches to dot-product similarity and full-parameter fine-tuning, enabling embedding directions and norms to jointly encode semantic information. We further introduce Focal Embedding Loss, a confidence-adaptive objective that downweights queries with high positive retrieval confidence while emphasizing ambiguous queries with competitive negatives. Experiments across multiple backbone scales and diverse multimodal embedding tasks consistently validate the effectiveness of PACE.

---

215. AdaVSkip: Adaptive Visual Token Skipping Across Layers For Efficient MLLMs Inference

• Authors: Yuyao Sun , Tao Deng , Shuang Li , Deqing Wang
• URL: https://arxiv.org/abs/2609.15131
• Abstract:

  Multimodal large language models (MLLMs) require substantial computation to process numerous visual tokens across all transformer layers. Most methods for efficient MLLM inference exploit horizontal redundancy by compressing visual tokens. Beyond token reduction, recent studies exploit vertical redundancy through early exit or fixed-layer skipping. However, we find that the extent and distribution of this redundancy vary across inputs and differ between self-attention and MLP modules. Motivated by these observations, we propose AdaVSkip, which equips each layer with two lightweight routers that independently determine whether visual tokens pass through by or skip the self-attention and MLP modules. These decisions collectively define an input-specific visual-computation path, but their discrete and non-differentiable nature makes learning effective paths challenging. To address this challenge, we develop a progressive two-stage training framework that updates only the routers while keeping the backbone frozen. Stage I establishes an initial routing policy through supervised training with input-specific targets derived from module-wise necessity scores. To further align the routing policy with task performance, Stage II uses reinforcement learning to optimize routing decisions with direct feedback from generated answers. It combines an answer correctness reward with a skip-consistency reward that discourages excessive retention of visual-token computation. Across three MLLM backbones, AdaVSkip maintains strong task performance with substantially less computation. On LLaVA-NeXT-7B, AdaVSkip reduces FLOPs by 53.2\% while preserving the original model’s average performance. Combining it with visual token compression increases this reduction to 91.2\%, while retaining 97.2\% of the original performance on average.

---

216. MoME: Mixture-of-Memory Embeddings for Context-Aware Sparse Lookup

• Authors: Muchen Li , Leonid Sigal , Renjie Liao
• URL: https://arxiv.org/abs/2609.15126
• Abstract:

  Scaling large language models efficiently has motivated sparse capacity mechanisms such as Mixture-of-Experts and, more recently, conditional memory: token-indexed embedding tables that augment the backbone with cheap parametric lookups. Existing memory-embedding methods retrieve via a deterministic function of the surface form, which collapses different contextual senses of the same token (e.g., python the language vs. the animal) into a single fixed entry. We introduce Mixture of Memory Embeddings (MoME), a context-aware memory mechanism that replaces each token’s single memory row with a mixture of M slots and uses a learned gate over the hidden state to choose which slots to read at each position. In controlled pretraining experiments across nanochat, Llama-3/MobileLLM, and Qwen3 backbones, MoME improves over Value Embedding, Bigram, and STEM baselines in iso-parameter and iso-training-FLOP settings, shows a more promising memory-size scaling trend at sub-billion scale, and remains efficient in training and inference. Qualitative routing analyses on polysemous tokens further suggest that the learned mixture exhibits a degree of semantic interpretability, dispatching the same surface token to distinct memory slots under different senses.

---

217. Refinement-based Flow Policy Optimization

• Authors: Bumgeun Park , Hyukjun Yang , Donghwan Lee
• URL: https://arxiv.org/abs/2609.15123
• Abstract:

  Flow-based policies offer an expressive representation for online reinforcement learning, but conventional flow matching requires samples drawn from the distribution to be modeled. This poses a challenge when the desired action distribution is defined only implicitly by a Q-function, since directly sampling actions from the resulting distribution is generally intractable. We propose Refinement-Based Flow Policy Optimization (RFPO), a novel framework for training a flow policy in online reinforcement learning by alternating between Q-guided sample refinement and self-target flow matching. RFPO first generates actions from Gaussian noise using the current flow policy and then uses a finite-step stochastic refinement procedure to move them toward an energy-based distribution induced by the Q-function. Each refined action is then paired with its corresponding initial noise sample and used as a fixed target for flow-matching training. By repeatedly refining its own outputs and learning from the resulting targets, RFPO incorporates Q-guidance into the policy without requiring direct samples from the target distribution, while retaining the capacity to represent multiple action modes. We further provide a theoretical analysis of the distributional dynamics induced by RFPO. Across six continuous-control tasks, RFPO matches or outperforms a standard Gaussian-policy baseline on almost every task. Experiments on six synthetic two-dimensional target distributions with diverse geometries demonstrate that RFPO captures complex multimodal structure without mode collapse.

---

218. DepthBenchCAD: When Does Deeper Auditing Yield More Reliable Conclusions?

• Authors: Hongye Yang , Zhihao Xie , Shengjun Xiong , Boxiao Huang
• URL: https://arxiv.org/abs/2609.15122
• Abstract:

  Generative CAD models are expected to remain behaviorally correct after parameter edits, so increasing the number of edit checks is often treated as a direct route to more reliable evaluation. Under a fixed budget, however, auditing each program more thoroughly reduces the number of tasks and independent generations that can be evaluated, which can ultimately make model-level estimates less accurate. We study this phenomenon and the conditions under which it arises. We decompose behavioral evaluation into three evidence levels: task templates, stochastic generations, and within-program edits. We define an average failure risk that is invariant to audit depth, and combine three-level variance with measured execution costs to analyze the tradeoff between deeper edit auditing and broader independent coverage. Experiments across two CAD environments and five generation systems show that the value of deeper auditing depends on where evaluation uncertainty originates. When template heterogeneity or generation stochasticity dominates, additional edit checks can increase total estimation error; when within-program state variation is large and generation is expensive, deeper auditing is more valuable. Variance and cost estimates from calibration predict the direction of this change and provide a diagnostic basis for allocating evidence on held-out tasks. These results show that the thoroughness of program inspection can diverge from the reliability of model evaluation, and they help determine whether the next unit of budget should be spent on a new task, a new generation, or additional edit checks.

---

219. Legislating World-Model-Based Planning with Legal Reasoning

• Authors: Dylan Waldner , Yiannis Kantaros , Guido Governatori , Risto Miikkulainen , Amir Banifatemi
• URL: https://arxiv.org/abs/2609.15113
• Abstract:

  As robotic systems grow more general, legal norms are needed to integrate them into society. This paper extends the isomorphism problem of aligning legal source texts with their encodings, and measures two key challenges to robot normative control: (1) the \textit{grounding isomorphism gap}, where perception error grounds false atoms for legal reasoning, and (2) the \textit{ontological isomorphism gap}, where one legal conclusion admits many faithful translations into planning constraints. The paper introduces a legal planning stack that employs Defeasible Deontic Logic (DDL) to constrain a motion planner. The stack leverages learned world models to plan and to provide legal context, enabling \textit{ex ante} governance that intervenes before an illegal action is executed. It was deployed on a simulated robot arm pushing a cube across a $3\times3$ grid. The findings were (1) the legislated agent abided substantially more often than the non-legislated one, and modeling perception uncertainty lifted abidance even further, (2) the legal reasoning ran efficiently at runtime and its verdicts were auditable, and (3) the stack adapted to exogenous signals and endogenous rule changes. Both gaps were measured: (4) world model and probe error corrupted the factual input for the DDL reasoner, and (5) a single law admitted several faithful metric interpretations yielding drastically different abidance. Thus, \textit{ex ante} legislation functions as intended, and closing these gaps with a standardized mapping from the law to runtime constraints and improved fact grounding from perception will yield robust laws that align robot behavior with society’s norms.

---

220. Physics Informed Neural Network model for the dynamical study of Abdominal Aortic Aneurysm

• Authors: Adrián Robles Arques , Martín Ruiz Fernandez , Javier Sanchis , Miguel A. Teruel , Juan Trujillo
• URL: https://arxiv.org/abs/2609.15104
• Abstract:

  We present the development and application of a three-dimensional Physics-Informed Neural Network (PINN) framework for the investigation of haemodynamic behaviour in the human aorta. The model incorporates a time-resolved simulation of pulsatile blood flow over a two-minute interval, enabling the extraction of pressure and velocity fields with high temporal fidelity. The mechanical stress exerted on the aortic wall was quantified through Laplace’s law, with temporal averaging applied to derive representative stress distributions. This approach circumvents the computational overhead associated with conventional computational fluid dynamics (CFD) methods by eliminating mesh generation and exploiting the automatic differentiation capabilities inherent to neural networks. The proposed methodology demonstrates that PINNs can serve as an efficient and accurate alternative for modelling complex vascular flow phenomena, offering significant advantages in scalability and computational cost reduction while maintaining physical consistency.

---

221. ChatGPT Images 2.5 in the Wild: A Launch-Period Dataset and Detector Evaluation

• Authors: Dennis Ng , Xingyu Shen , Ankit Raj , Kidus Zewde , Tommy Duong , Yuchen Zhou , Yuxin Zhang , Neo Tiangratanakul , Simiao Ren
• URL: https://arxiv.org/abs/2609.15100
• Abstract:

  An image tool can change its underlying generator while retaining its public name, making version attribution from online posts ambiguous. We study this problem after the ChatGPT Images 2.5 launch. Our frozen collection contains 3,478 images from 2,440 posts across 8 sources. Recorded posting times fall within the first 51.1 hours after the announcement. It records three attribution tiers and retains standalone images after image-form filtering and targeted review. Caption claims and host records provide admission evidence, not independently verified generator identity. The observed content profile depends on the source mixture: NightCafe supplies 39.0% of images but 77.0% of CLIP-assigned fantasy scenes. We then evaluate six frozen detectors at thresholds calibrated to a 5% flag rate on reference photographs. Collection flag rates range from 3.7 to 56.4%, falling 42-81 percentage points below GenImage recall. Held-out artwork false-positive rates range from 1.5 to 96.5%, so a higher collection flag rate does not by itself establish better detection. An exploratory X-only comparison with our April collection finds a higher September flag rate for Effort, and a suggestive difference for DoU, under fixed-threshold post-clustered bootstrap intervals. Attribution, content and processing differences prevent a causal interpretation of these contrasts. The collection supports analysis of reported model use during a product transition, with source and attribution evidence retained for interpretation.

---

222. Generate to Explore, Select to Exploit: Aligning LLM-based Headline Generation with Personalized Recommendation

• Authors: Yi Chen , Rufeng Cheng , Qiang Xie , Tao Li
• URL: https://arxiv.org/abs/2609.15094
• Abstract:

  In industrial recommendation feeds, presenting a static headline for an item often fails to satisfy the diverse, multimodal interests of the user population, particularly suppressing the needs of long-tail audiences. While Large Language Models (LLMs) have been integrated into recommendation for content understanding or ranking, directly optimizing them to output a single best headline typically leads to mode collapse—converging to generic patterns that satisfy average tastes but miss specific latent intents. To bridge this gap, we introduce GESE (Generate to Explore, Select to Exploit), a framework operating at the system’s presentation layer that decouples personalization into generative exploration and selective exploitation. First, we treat the LLM as a probabilistic explorer, utilizing Group Sequence Policy Optimization (GSPO) with a hierarchical reward mechanism to generate a candidate set that maximizes the semantic coverage of potential user interests. Subsequently, a lightweight, real-time feedback-aware selector acts as the exploiter, identifying the optimal realization from the candidate pool based on instant contextual signals. Extensive deployment on a commercial platform with over 100 million daily active users demonstrates that GESE significantly outperforms state-of-the-art baselines, achieving a 2.57% lift in CTR and 0.87% in dwell time. These results validate that decoupling diversity-oriented generation from precision-oriented selection offers a robust blueprint for aligning generative AI with dynamic user utility.

---

223. Beyond Numerical Time Series: A Unified Benchmark for Multimodal Forecasting with Heterogeneous Context

• Authors: Peng Chen , Zhihao Zhuang , Hongzhou Chen , Junhao Huang , Aiping Yang , Mengsen Wu , Yiding Liu , Xilin Dai , Zewei Dong
• URL: https://arxiv.org/abs/2609.15087
• Abstract:

  Most time series forecasting benchmarks remain numerical-centric and provide limited support for evaluating contextual information that shapes real-world temporal dynamics. Existing multimodal benchmarks also suffer from limited data and context coverage, fragmented evaluation settings, and overreliance on aggregate evaluation. In this paper, we propose \textbf{MUSE-Bench}, a unified benchmark for multimodal time series forecasting with heterogeneous context. It comprises fourteen datasets across eight domains and six types of context: metadata, events, holidays, news, images, and numerical covariates. We evaluate diverse forecasting paradigms, including statistical, data-specific, foundation, multimodal, and general-purpose LLM forecasting methods under shared non-overlapping forecast windows, common target observations, and consistent point and probabilistic metrics. Extensive experiments yield three main findings. First, numerical time series foundation models dominate the overall ranking, while Aurora, the evaluated multimodal foundation model, trails the leading numerical TSFMs but outperforms all evaluated data-specific models. Second, ablations show that external context improves the four evaluated context-aware models, whereas incorrect or temporally misaligned context degrades performance. Third, general-purpose LLMs perform poorly as direct forecasters, and LLM-guided refinement does not yield consistent improvements. MUSE-Bench enables systematic evaluation of how forecasting models utilize context and provides a foundation for future multimodal forecasting research.

---

224. Translating the Translator: Decomposing the Cost of English-Forced Inter-Agent Communication

• Authors: Kushagra Agrawal , Yuming Feng , Man-Fai Leung
• URL: https://arxiv.org/abs/2609.15079
• Abstract:

  Multi-agent LLM architectures, such as LangChain and AutoGen, largely assume English as the lingua franca for internal inter-agent communication, even when the end-user task is non-English. We fill this gap by evaluating a two-agent extraction-answer core, with an additional back-translation agent in the English-forced condition, across four typologically diverse languages (Hindi, Chinese, Spanish, Arabic; n = 300 per language) using the Aya-23-8B model. We compare a native-language pipeline to an English-forced one (which incorporates a final back-translation step from English to the user’s language). We discover a statistically significant English-Forcing Tax (surviving a strict Bonferroni correction) that isolates the cost of English routing from general multi-agent orchestration overhead. Forcing inter-agent communication through English reduces Exact Match accuracy by 13.0 percentage points (Spanish) up to 30.6 percentage points (Hindi) compared to native-language multi-agent execution. Using chrF scores as a diagnostic measure of English-reference lexical overlap, we find that lower overlap is strongly associated with pipeline failure, consistent with translation loss being an important contributor to the observed performance drop. These findings suggest a compelling case for native-language routing in agent frameworks when the source and target languages are typologically distant, reducing a compounding translation tax.

---

225. Branched Optimal Transport Amortization

• Authors: Semyon Semenov , Viktor Kovalchuk , Meir Roketlishvili , Albert Baichorov , Fakhri Karray , Martin Takac , Arip Asadulaev
• URL: https://arxiv.org/abs/2609.15072
• Abstract:

  Methods of Branched Optimal Transport (BOT) mimic the economy and efficiency of natural tree-like structures, such as those found in rivers and biological systems. These methods are widely applicable for designing efficient networks in society, from river basins and blood vessels to mail and gas distribution systems. However, they remain understudied in the context of designing deep generative models, particularly at a large scale. Standard continuous-time generative models, such as the flow matching approach, fail to capture the inherent hierarchical and branching patterns present in real-world data. Current models provide no mechanism for flows to merge or share pathways to minimize total transport cost. Inspired by the “economy of scale” principle in BOT, we introduce a novel, scalable branched flow-matching algorithm designed to solve the branched optimal transport problem in high dimensions. Our method adapts the Benamou-Brenier continuous-time optimal transport formulation to learn branched generative flows. These flows allow probability mass to aggregate along common pathways before branching out to diverse targets. Parametrized by neural networks, our method effectively learns complex branched generative processes. We demonstrate its effectiveness on challenging high-dimensional tasks in biology and image generation.

---

226. Rethinking Procedural Audio Pre-training: Source Scaling and Objective Adaptation

• Authors: Jiajun Peng , Fengrui Liu , Xinyu Liu , Feng Liu
• URL: https://arxiv.org/abs/2609.15067
• Abstract:

  Procedural audio has emerged as a viable source for transferable audio representation learning, but its design principles remain this http URL revisit two questions: how a procedural source should be scaled, and whether training choices developed on natural audio should transfer unchanged to procedural this http URL a controlled source, we separate scale into formula-class coverage C and within-class rendering diversity this http URL with FDSL and AudioMAE show that these two forms of scale provide different benefits and depend on the learning formulation and downstream task. A matched AudioMAE study further shows that procedural audio favors low mask ratios (10%–25%), whereas AudioSet-28K favors 50%–75%. Shared-codebook analysis reveals lower patch diversity and stronger temporal predictability in procedural audio. These results motivate source-aware procedural pre-training, where source scaling and learning configuration are considered this http URL is available at this https URL .

---

227. Salesforce Koa: An Enterprise Language Model for Agentic Tool Use

• Authors: Zixiang Chen , Sufeng Niu , Yingchi Liu , Wenting Zhao , Akshara Prabhakar , Shubham Mehrotra , Bin Bi , Zhujun Lan , Katherine Tan , Mohammad Ramezanali , Tulika Manoj Awalgaonkar , Monojit Banerjee , Jielin Qiu , Shiva Kumar Pentyala , Zhepeng Cen , Anupam Tripathi , Ali Ziaei , Regunathan Radhakrishnan , Darvish Lee Shadravan , Shelby Heinecke , Sitaram Asur , Silvio Savarese , James Zhu , Phil Mui , Huan Wang
• URL: https://arxiv.org/abs/2609.15066
• Abstract:

  We present Salesforce Koa, an enterprise language model built by post-training the open-weight Nemotron-3-Super-120B foundation model with reinforcement learning using Group Relative Policy Optimization (GRPO). Salesforce Koa is trained on public and synthetically generated data, with no customer data, to improve tool use and agentic capabilities while preserving strong general-purpose performance. Its distinctive component is a simulation-to-reward pipeline that expands workflow specifications into persona-conditioned multi-turn tasks with task-resolution rewards grounded in successful tool use for data-dependent requests. For enterprise domains, these specifications are written in Agent Script, Salesforce’s declarative language for building Agentforce agents; for public tool-use domains, we synthesize the workflow structure directly. The same simulation and grounded-reward machinery drives GRPO across both. Across public tool-use, agentic-reasoning, and enterprise Customer Relationship Management (CRM) benchmarks, Salesforce Koa improves over its open-weight base, with the clearest gains on multi-turn tool use, and surpasses a strong proprietary baseline while remaining below the strongest frontier models. These results show that specification-driven reinforcement learning is a practical path to specializing open-weight foundation models for enterprise agentic tasks.

---

228. Not All Prompts Are Equal: Exploration-Guided Prompt Scaffolding for Multimodal Reinforcement Post-Training

• Authors: Yuanhao Yue , Qianli Ma , Chengyu Wang , Haoting Wang , Lei Shen , Jun Huang
• URL: https://arxiv.org/abs/2609.15051
• Abstract:

  Training prompts in online reinforcement learning (RL) differ substantially in how informative they are for the current policy: some are already saturated while others are too difficult to yield reliable learning signals, yet both receive equal rollout budget under standard training. We propose an exploration-guided prompt scaffolding framework that adapts the training prompt distribution dynamically throughout RL post-training of multimodal large language models (MLLMs). Central to our approach is the $\textit{Exploration Potential Score} (EPS)$, a lightweight rollout-based proxy for prompt utility derived from KL-regularized policy improvement theory, computable directly from on-policy rollout statistics without additional overhead. Rather than discarding low-utility prompts, we use a teacher model to generate scaffolded rewrites that preserve the original task intent while making subsequent training more informative, reframing teacher supervision as training-data refinement rather than output imitation. Integrated with GRPO on Geo3K and MMK12, our method consistently outperforms the baseline on both in-domain and out-of-distribution benchmarks, achieving up to 9.7\% relative improvement in-domain and gains of 11.5\% on MathVision and 11.1\% on MMMU-Pro.

---

229. Ensemble Complexity in Photovoltaic Forecasting

• Authors: Sun Ze , Zhou Liguo , Xu Yuqing , Yu Lei , Jiang Mingming
• URL: https://arxiv.org/abs/2609.15049
• Abstract:

  An ensemble can improve photovoltaic forecasts while adding components that contribute little or increase computation. We assess these effects through matched comparisons and ablations of a fixed heterogeneous predictor bank. Hourly experiments use GEFCom2014 and three additional public datasets, with chronological partitions and three seeds. Under retrospective ERA5 assistance, static fusion reduces scaled mean absolute error against matched boosting by 1.11%, 4.41%, and 1.63% on PVDAQ, OPSD, and Ausgrid; only OPSD remains supported after multiple-comparison correction. Weather gating offers no consistent incremental benefit. Exploratory member removals show group-level dependence alongside individual redundancy. A separate, previously inspected fifteen-minute case replaces one neural member with a tree predictor: normalized error falls by 1.72%, but measured inference is slower. These findings support component-wise evaluation with explicit limits on weather availability and test-set reuse.

---

230. Personalizing Personal Health Interfaces: Co-Design with Generative AI

• Authors: Karthik S. Bhat , Vidhi Shah , Vedika Agnihotri , Dong Whi Yoo , Koustuv Saha
• URL: https://arxiv.org/abs/2609.15046
• Abstract:

  Personal health interfaces present wellbeing data through standardized dashboards that rarely fit how people interpret or act on it. Personalizing them to what people would like to see for themselves often requires design and technical expertise, a barrier that generative AI may potentially lower. Therefore, we ask what designs emerge and how it enables and constrains the design process. We conducted a co-design study where 14 participants redesigned Google and Apple Health interfaces using Figma Make. Participants reimagined interfaces that supported personal context, future planning, and interactive experiences, yet conversational AI designs converged around chat-window conventions. AI helped materialize loosely articulated ideas, but model defaults and generation latency shaped iteration. The process more readily operationalized interpretability and accountability than privacy, trust, and emotional safety. Generative co-design let participants create interfaces directly, blurring the boundary between intentions and model defaults. We discuss implications for preserving agency and flexible user-directed interfaces.

---

231. Mirror, Mirror on the Wall: Prompt Echoing in Small Instruct Language Models

• Authors: Inez Okulska , Bartosz Naskręcki , Jan Piotrowski , Tomasz Steifer
• URL: https://arxiv.org/abs/2609.15045
• Abstract:

  Prompt echoing is a recognized failure mode of instruct language models, in which a model instead of generating a response, mirrors the provided prompt, even though it did not receive a specific instruction to do so. Is this phenomenon a sign of the model leaking the content of its training dataset, or is it rather caused by a misaligned behavior of the internal induction/copying mechanisms? We investigate prompt echoing small language models from different families (Gemma, Llama, Qwen, SmolLM and OLMo) and show that echoing prompts are likely to have partial overlap with the training dataset but the phenomenon is primarily driven by the model’s induction heads.

---

232. SpliTEE: Improving LLM Inference on Trusted Hardware with Differentially Private GPU Outsourcing

• Authors: Shashie Dilhara Batan Arachchige , Robin Carpentier , Hassan Jameel Asghar , Dali Kaafar
• URL: https://arxiv.org/abs/2609.15039
• Abstract:

  User prompts provided to large language models (LLMs) may contain sensitive or private information that can be misused by remotely deployed models, such as through inadvertent memorization during retraining. One way to protect user prompts is to execute the LLM inside a trusted execution environment (TEE), with the guarantee that the service provider has no access to computations performed within or information exchanged with the TEE. However, current TEEs are primarily CPU-based and significantly slower than GPUs optimized for LLM inference. To circumvent this, Tramer and Boneh (2019) proposed Slalom, which splits neural network inference between a TEE and an untrusted GPU and encrypts intermediate inputs sent to the GPU. We extend this split-inference architecture to LLM inference and instead protect intermediate inputs using differential privacy. We show that masking intermediate representations is necessary by showing that a prompt-reconstruction attack can recover prompts from these representations with nearly 80% accuracy. Our main contribution is a global sensitivity analysis of key LLM functions, which bounds the required scale of differentially private noise. Unlike encryption, differential privacy avoids quantization, allowing the LLM to remain in the floating-point domain. We also derive an upper bound on floating-point error from masking and noise cancellation in the TEE as a function of the privacy parameter epsilon. We implement our architecture using Intel TDX and evaluate it with two LLMs: Llama-3.2-3B and Qwen3-4B. Our split execution is nearly twice as fast as fully CPU-based inference inside TDX and 5-15 seconds faster than encryption-based Slalom while achieving higher accuracy. Finally, we demonstrate that prompt reconstruction, even with knowledge of the differential privacy mechanism, cannot recover more information than is contained in an unrelated prompt.

---

233. Validating Hybrid-State Cache Recovery for GLM-5.3-Flash with vLLM and LMCache

• Authors: Frank Li
• URL: https://arxiv.org/abs/2609.15030
• Abstract:

  External cache transfers can succeed while a hybrid language model resumes from an inconsistent state. We examine the full 45-layer GLM-5.3-Flash model, using the RedHatAI/ GLM-5.3-Flash-NVFP4 quantized checkpoint with vLLM and LMCache under four-way tensor parallelism. A complete-hit recovery mismatch restored state for the full prompt while the scheduler credited one fewer token. We aligned recovery through strict-prefix lookup and established a numerical comparison using shared computation corrections, matched checkpoint scheduling, and fixed per-rank kernel configurations. In a nine-length serial workload, agreement with the modified recomputation control improved from 34/36 to 36/36 generations, each containing 64 token IDs. A separate instrumented run passed recorded transfer-page, effective-tail, and delayed-save checks. Three additional synthetic templates passed 72 paired 256-token continuations across two fresh-container runs. A subsequent serial performance study preserved output equality across 120 requests; among the measured trials, CPU reload reduced time to first token by 46-64% and total request time by 1.9-7.0% relative to modified cold recomputation. The contribution is an experimentally validated integration repair applying an existing checkpoint-alignment principle. The evidence is confined to one model revision and controlled configuration; it does not establish general determinism, task-quality equivalence, concurrent-serving gains, or capacity beyond GPU memory.

---

234. Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks

• Authors: Aashiq Muhamed , Mona T. Diab , Virginia Smith , Andrew Ilyas , Matthew Jagielski
• URL: https://arxiv.org/abs/2609.15029
• Abstract:

  Backdoor poisoning attacks add poisoned examples to otherwise-clean finetuning data, pairing a trigger with a target behavior that the model learns to produce when the trigger appears. Existing evaluations typically fix the number of poisoned examples and sample them at random from a candidate pool. We show that this can severely underestimate worst-case vulnerability: across three LLaMA-3-8B backdoor settings, holding the model, clean data, and poison count fixed, attack success ranges from 3% to 80% depending only on which poison set is chosen. We formalize poison selection as oracle-budgeted set optimization and introduce SAILS (Set-level Audit-Informed Iterative Learned Selection), which learns a set scorer from a few hundred finetune-and-evaluate runs, ranks millions of candidate sets, and audits only a small shortlist. SAILS improves held-out attack success by 30 percentage points on average over the strongest influence baselines, transfers from small-scale to full-scale finetuning, and extends to code-generation, agentic, and API-only backdoors.

---

235. PIDS-Bench: Evaluating Prompt-Injection Detectors Under Over-Defense, Obfuscation, and Distribution Shift

• Authors: Yusuf Khalid Shire , Sang-Chul Kim
• URL: https://arxiv.org/abs/2609.15017
• Abstract:

  Prompt-injection detectors are typically evaluated using aggregate F1 on in-distribution test data, which offers limited insight into behavior under distribution shift, particularly on the benign side of the decision boundary, where false positives impose direct operational cost yet are seldom measured. We present PIDS-Bench, a frozen multi-axis benchmark that jointly evaluates attack detection and benign false-positive behavior at fixed thresholds, spanning in-distribution inputs, hard-benign prompts that mimic injection structure without malicious intent, obfuscated attacks, and domain and structural distribution shifts. We evaluate seven detectors (learned baselines, external prompt-injection classifiers, and broad-safety comparators) alongside a rule-based lower-bound reference. Multi-axis evaluation exposes a failure mode that aggregate F1 conceals. A detector exceeding F1 = 0.98 on the held-out split still misclassifies roughly one-third of an externally-sourced benign subset drawn from public corpora and restricted to security-adjacent content. Across a full threshold sweep and five training seeds, no internal detector reaches an operating point satisfying F1 >= 0.95 and hard-benign FPR <= 0.10 together on this stress distribution. Decomposing by provenance, we find that hard-negative augmentation nearly eliminates over-defense on curated stress inputs but leaves it substantially intact on externally-sourced prompts, a pattern we term provenance-sensitive over-defense. The asymmetry holds across both fine-tuned architectures and does not diminish as the augmentation pool grows, with the externally-sourced FPR remaining far above the 0.10 target. Whether augmentation matched to the externally-sourced distribution would close this gap is untested; threshold calibration and curated-style augmentation alone do not.

---

236. Steering Generative Robot Policies with Lexicographic Preferences

• Authors: Yixuan Jia , Jonathan P. How
• URL: https://arxiv.org/abs/2609.15014
• Abstract:

  Pretrained generative robot policies can produce effective behaviors across diverse environments, but deployment can lead to requirements and preferences that may not have been represented during training. Furthermore, at deployment, an operator, user, or application may assign these requirements and preferences a priority order that can vary across deployments. For example, embodiment-specific feasibility constraints may need to be satisfied first, while user-specific preferences guide behavior among the feasible options. We show that a frozen generative robot policy—based on either diffusion or flow matching—can be steered at inference time to respect such lexicographically ordered deployment objectives. To achieve this, we introduce two modifications to the sampler. First, we apply dynamic-barrier guidance to sampled trajectories, constraining lower-priority updates so that higher-priority costs do not increase (up to first order). Second, we select the executed sample using a cascade that successively filters candidate samples according to each priority level. The policy weights remain unchanged. On a navigation benchmark, we demonstrate that our method improves success, traversability, and preference compliance over the frozen policy, and achieves substantially better compliance than tuned weighted-sum baselines. The same method transfers to a flow-matching manipulation policy on LIBERO, where it improves compliance without reducing task success. A controlled manipulation study further shows that, in settings where a fixed weight can match the desired ordering, the dynamic barrier reaches comparable best performance over a substantially wider range of parameter settings.

---

237. IMPACT-VLA: Interaction-aware Multimodal Propagation Attribution via Counterfactual Trajectories for Vision-Language-Action Policies

• Authors: Jinwoong Kim , Sangjin Park
• URL: https://arxiv.org/abs/2609.15005
• Abstract:

  Vision-Language-Action (VLA) policies perform robot manipulation tasks using multimodal inputs such as visual observations, proprioceptive states, and language instructions. However, it remains unclear at which execution stages each modality contributes to final task success and how input interventions propagate through subsequent states, observations, and actions. Existing attribution approaches primarily measure local sensitivity or temporally aggregated importance, limiting their ability to capture phase-dependent contributions and cross-phase dependencies. We propose Interaction-aware Multimodal Propagation Attribution via Counterfactual Trajectories for Vision-Language-Action Policies (IMPACT-VLA). IMPACT-VLA constructs behavioral phases from action transitions in a successful reference rollout, aligns them with policy query boundaries, and defines phase-modality blocks as attribution units. It then performs closed-loop counterfactual re-execution to quantify each block’s contribution to final task success. We further analyze cross-phase non-additive interactions and trajectory propagation while distinguishing behavioral from functional recovery. Across 30 LIBERO robot manipulation tasks using OpenVLA-OFT, dominant-modality transitions occurred in 25 tasks (83.3%), and closed-loop attribution identified task-critical information more faithfully than Static Action Perturbation. Later-block marginal gains for negatively interacting pairs increased by approximately 3.3x under early-phase input replacement, while functional recovery could occur without behavioral recovery. These results reveal when multimodal inputs support task success and how their contributions become conditionally coupled during closed-loop execution.

---

238. ActGuard: Pre-execution Action Auditing against Indirect Prompt Injection in LLM Agents

• Authors: Bingzheng Wang , Xiaoyan Gu , Wentao Wang , Xingyou Yang , Hongcheng Li , Rong Yin
• URL: https://arxiv.org/abs/2609.14987
• Abstract:

  Large language model (LLM) agents interact with external environments through tool invocation, but tool outputs can also expose them to indirect prompt injection (IPI) attacks. Existing defenses mainly rely on prompt hardening, content filtering, pre-generated plans, or permission constraints. These approaches often struggle with complex tasks or over-sanitize external content, making it difficult to balance security and utility. The key challenge is therefore to preserve execution flexibility while precisely identifying and removing the malicious content that actually induces unsafe actions. To address this challenge, we propose ActGuard, a pre-execution action auditing framework. Rather than judging whether external content is inherently suspicious, ActGuard assesses whether it causes the current action to deviate from a locally reasonable expectation. At each step, ActGuard predicts the tools likely to be used by the upcoming action and constructs a local tool prior without constraining the execution trajectory. Before execution, it compares the candidate action against this prior and performs tool-level contrastive analysis and parameter-level evidence localization to identify deviations in tool selection and action parameters. A verifier then examines the localized evidence, masks only spans confirmed as malicious, and regenerates the action from the sanitized context. This design preserves legitimate planning flexibility while minimizing information loss from indiscriminate filtering. We evaluate ActGuard on challenging benchmarks for tool-using agents. Results show that ActGuard reduces attack success rates to a level comparable to state-of-the-art defenses while maintaining task utility close to the no-attack setting, achieving a favorable security-utility trade-off. Our code is publicly available at: this https URL .

---

239. LiftGCN: Efficient Energy-Preserving Graph Learning via Joukowski Spectral Lifting for Finite Element Stress Prediction

• Authors: Chen Zeng , Qiao Wang
• URL: https://arxiv.org/abs/2609.14977
• Abstract:

  Finite element stress fields often exhibit strong local non-smoothness, where stress concentrations near holes, notches, and loading regions induce sharp spatial gradients and high-frequency graph components. Although graph neural networks naturally operate on irregular finite element meshes, conventional message passing is inherently smoothing and progressively attenuates such high-frequency information. Unitary propagation alleviates this problem by preserving spectral magnitudes, but typically relies on matrix functions and high-order approximations with $O(Ked)$ propagation complexity. We propose LiftGCN, an efficient spectrally stable graph network based on Joukowski spectral lifting. LiftGCN maps the real spectrum of a normalized graph operator onto the unit circle through the Joukowski relation and realizes the resulting spectral transformation as a simple second-order recurrence, avoiding matrix exponentials, eigendecomposition, and high-order polynomial truncation. We show that the linear Joukowski backbone has unit-modulus characteristic roots and admits an energy-preserving structure under a positive-definite metric, preventing exponential attenuation of graph-frequency components with depth. Each layer requires only one sparse neighborhood aggregation, yielding $O(ed)$ propagation complexity, while lightweight local nonlinear residuals provide expressive feature transformations. Experiments on finite element stress prediction demonstrate that LiftGCN achieves competitive overall accuracy while improving reconstruction of stress concentrations and local high-gradient structures with substantially reduced computational cost. Our code is available at this https URL .

---

240. Online Language Adaptive Sampling for Better Distributed Cross-lingual Gains

• Authors: Quang Phuoc Nguyen , Félix Gaschi , David Anugraha , Santiago Martínez Novoa , En-Shiun Annie Lee
• URL: https://arxiv.org/abs/2609.14969
• Abstract:

  Realignment is a promising approach for improving the cross-lingual transfer ability of multilingual language models, particularly for extremely low-resource languages (LRLs). However, existing realignment methods rely on uniform and random sampling of parallel sentences across languages, which may be suboptimal under limited batch sizes. In practice, models may benefit from seeing certain languages more frequently, especially those that are poorly aligned, and the optimal distribution can evolve throughout training. In this work, we propose a simple yet effective adaptive sampling strategy that assigns trainable sampling probabilities to each language. Languages that contribute more to the realignment loss are sampled more frequently in subsequent batches, and the optimal distribution can evolve throughout training. Our method employs an inner-outer optimization loop with a small overhead, leading to consistent performance improvements and, more importantly, distributing the gains across languages. We observed a $+0.67$ average performance increase on all tasks with XLM-R, and $+0.60$ with Gemma 2 9B compared with uniform realignment. Furthermore, our method is robust across different models. Code available at this https URL .

---

241. CAL-MOS: Bridging Layers with Adapters for Robust MOS Prediction Across Speech Foundation Models

• Authors: Alef Iury Siqueira Ferreira , Pedro Lustosa Rege Botelho , Fernanda Silva , Daniel Casanova , Rafael Faustino , Frederico Oliveira , Arlindo Galvão Filho , Anderson da Silva Soares
• URL: https://arxiv.org/abs/2609.14956
• Abstract:

  Speech Quality Assessment (SQA) is essential for modern speech technologies, and recent non-intrusive SQA predictors increasingly rely on Speech Foundation Models (SFMs). However, because SFMs expose representations from many layers, it remains unclear which depths are most informative for MOS prediction and how multi-layer information should be combined reliably across backbones and datasets. We benchmark ten SFMs on four MOS datasets under three regimes: full fine-tuning, last-layer probing with a frozen encoder, and naive cross-layer weighted aggregation. We find that the best layer is strongly backbone- and dataset-dependent, and that naive weighted fusion can be unstable across settings. We further evaluate a layer-calibrated aggregation variant that applies per-layer adapters before pooling, which improves the robustness of multi-layer fusion and narrows the gap to full fine-tuning while keeping the backbone frozen.

---

242. Cross-Block Conditioning in Deep Boltzmann Machines for Statistical Data Fusion

• Authors: Junichiro Niimi
• URL: https://arxiv.org/abs/2609.14934
• Abstract:

  Statistical data fusion combines two panels that share a block of covariates but observe disjoint outcome blocks, and in its traditional form no row observes both outcomes at once. That rules out the discriminative criterion one would rather train a Deep Boltzmann Machine with, since multi-prediction training needs ground truth for whatever it holds out. We propose observed-block multi-prediction, which restricts the multi-prediction objective to targets drawn from what each row actually observes. It is well defined for any missingness pattern and reduces to the original criterion when rows are complete. Having a discriminative criterion that survives the setting lets us ask whether the joint model is needed at all, by separating what it contributes into a representation part and an inference part. On two consumer panels, on grids over sample size and covariate width spanning 35 cells and 875 runs, the fine-tuned DBM is the best of fifteen methods in every cell; but almost none of that advantage comes from generative pre-training, which is confined to the smallest sample size on one dataset and absent on the other. It comes from conditioning on one outcome block when predicting the other. This term amounts to +0.19 and +0.07 percentage points, is positive in all 35 cells, and, unlike every other contribution we measure, neither decays as the panels grow, nor requires a second hidden layer, nor requires more inference. Permuting one outcome block to destroy its association with the other removes the gain entirely, which is what the account predicts. The margins are small. But a small effect that does not decay is a different object from one that does, because it rests on evidence that no model mapping covariates to outcomes can accept.

---

243. Neural-Network Solutions to Real-Space Charge Density and Generalization

• Authors: Yuxuan Zeng , Taoyuze Lv , Zhicheng Zhong
• URL: https://arxiv.org/abs/2609.14906
• Abstract:

  The Hohenberg-Kohn theorem establishes that, in principle, the ground state (GS) charge density contains all GS information of a many-electron system, such that all GS observables can be expressed as functionals of the GS charge density. Conventional Kohn-Sham density functional theory requires iterative solution of the self-consistent-field equations at substantial computational cost, motivating the development of deep learning surrogates for electronic structure calculations and, in turn, accelerating computer-aided materials design. Here, we propose \textbf{AIDEN}, an \underline{A}tomic-\underline{I}nteraction \underline{D}ensity \underline{E}quivariant \underline{N}etwork for solving real-space charge density. AIDEN separates the element-dependent one-center density from environment-induced density redistribution and represents the latter through complementary atom- and edge-centered tensor correlations. A continuous low-rank Gaussian decoder then reconstructs the density at arbitrary spatial coordinates while reusing atomic encodings independently of the evaluation grid. AIDEN achieves state-of-the-art accuracy on periodic crystal benchmarks while remaining competitive for molecular systems, and further demonstrates zero-shot transferability across several structurally distinct out-of-distribution case studies. Furthermore, AIDEN provides substantially faster inference than both baseline models and full SCF calculations, enabling efficient charge density reconstruction for large-scale electronic structure calculations.

---

244. Forty Shades of Blue: Quality-Diversity Alignment via Mode-Conditioned Reinforcement Learning

• Authors: Jiayi Yuan , Hangoo Kang , James Jihao Liu , Yejin Choi , Vikram Iyer , Liwei Jiang , Natasha Jaques
• URL: https://arxiv.org/abs/2609.14896
• Abstract:

  A notable byproduct of LLM alignment training is mode collapse: the progressive loss of output diversity that narrows a model’s expressivity at inference time. This degradation is especially limiting for applications requiring open-ended exploration and pluralistic perspectives, such as scientific ideation and creative writing. We present MoDA (Mode-conditioned Diversity Alignment), an online post-training RL algorithm that jointly optimizes generation quality and diversity, inspired by the coordination perspective in multi-agent reinforcement learning (MARL). MoDA trains a single shared LLM policy conditioned on abstract numbered roles, where each role acts as an agent competing to produce outputs distinct from the others. This formulation encourages mode-conditioned agents to explore complementary regions of the high-quality output space without requiring hand-crafted personas or architectural modifications. MoDA employs a prompt-adaptive quality gating mechanism that calibrates a reference quality threshold and grants diversity rewards only to responses that meet the threshold, preventing reward-hacking behaviors that compromise response quality. To study quality-diversity tradeoffs, we evaluate MoDA on a comprehensive suite of benchmarks spanning seven general capability tasks and four domain-specific diversity tasks in scientific ideation and creative writing. MoDA improves SBERT diversity by 265% on the Infinite-Chat held-out prompts, while increasing average general capability pass@1 by 10.3% over the Qwen3-8B baseline. Compared with the strongest DivPO baseline, MoDA improves SBERT diversity from 0.274 to 0.482 (+75.9%) and E-Vendi from 2.86 to 4.4 (+53.8%), while improving average general capability pass@1 by 7.0%. Overall, MoDA provides a drop-in alternative to standard post-training methods that preserves and expands the model’s expressive output space while improving quality.

---

245. PeerPen: AI-Assisted Writing for Online Mental Health Peer Support

• Authors: Jiwon Kim , Sherry Gong , Maya Ajit , Soorya Ram Shimgekar , Yunhao Yuan , Dong Whi Yoo , Eshwar Chandrasekharan , Koustuv Saha
• URL: https://arxiv.org/abs/2609.14886
• Abstract:

  Online mental health communities thrive on peer support, yet those who volunteer to help often lack formal training and may struggle to articulate supportive responses. AI co-writing could lower this barrier; however, peer support derives much of its value from being perceived as personal, raising questions around authorship, ownership, and trust. We built PeerPen, a writing assistance tool embedded within a Reddit-like interface, supporting two main features: draft generation and revision of user-written responses. Through semi-structured interviews with 15 participants, we find that PeerPen reduced the burden of composing responses and increased confidence in offering support. Participants wanted AI to assist their writing without taking over authorship and anticipated tensions around authenticity and trust. Such assistance could make authorship uncertain even for responses written without it, weakening trust across the community. We contribute design implications for AI writing assistance that scaffolds supportive communication, preserves authorship, and accounts for community-level trust.

---

246. SeqMaestro: From nucleotide sequences to biological hypotheses through interpretable machine learning

• Authors: Evgeny S. Saveliev , Krzysztof Kacprzyk , Charlotte Capitanchik , Neelanjan Mukherjee , Kate Matlin , Ryan Sheridan , Srinivas Ramachandran , Jernej Ule , David L. Bentley , Mihaela van der Schaar
• URL: https://arxiv.org/abs/2609.14882
• Abstract:

  Nucleotide sequence analysis is central to problems spanning regulatory genomics, evolutionary biology, and phenotype prediction. Classical bioinformatics methods extract interpretable sequence properties such as motifs and k-mer composition, but their flexibility is limited. In contrast, modern deep learning models can learn powerful predictive representations directly from raw sequences, yet their internal representations and decision mechanisms are difficult to inspect. Interpretable machine learning methods (e.g., sparse linear models and decision trees) provide human-understandable representations of predictive relationships but are not designed to operate directly on nucleotide sequences. Here, we introduce SeqMaestro, a machine learning framework that proposes biological hypotheses from nucleotide sequences using interpretable models. Our solution is centered around a two-layer interface that connects nucleotide sequences with the broader ecosystem of interpretable machine learning. SeqMaestro uses this interface to fit diverse combinations of interpretable models, feature representations, and extraction strategies, leveraging variability across transparent models to identify robust biological signals and richer predictive relationships than feature importance alone can provide. The system also supports data transformation and cleaning, model fitting, hyperparameter tuning, reliability analysis, and synthesis of results into a contextualized written report. By providing these capabilities through a no-code workflow, SeqMaestro is designed to make interpretable sequence analysis accessible to researchers without requiring extensive programming or machine learning expertise. SeqMaestro thereby provides an accessible route from nucleotide sequences to biological hypotheses.

---

247. Interpolation Is Not Invariance: Pair Count Is Not Coverage in Transformation Audits

• Authors: Mohammed Ahnouch , Lotfi Elaachak
• URL: https://arxiv.org/abs/2609.14870
• Abstract:

  Counting equivalent pairs is a common way to report transformation-audit coverage, but it can substantially overstate the constraints imposed by an audit: pairs generated from the same semantic object are correlated, and complete orbit graphs contain algebraically redundant edges. We therefore distinguish four complementary quantities—edge count $m$, effective contrast rank $s$, population support rank $r$, and graph spectral gap $\eta$—and characterize their roles in audit coverage and deployment reliability. Under a rank-$r$ Gaussian contrast model, a population-invariant calibrated reader exists exactly when the anchor has a component in $\ker T$. When an audit has rank $s < r$, its unobserved risk is $R^\star/U$, with $U \sim \operatorname{Beta}((r-s)/2,s/2)$; when $s \ge r$, exact calibrated interpolation is infeasible. The same distinction appears in orbit topology: a spanning tree imposes the same exact-null constraints as a complete graph, while a sharp graph Poincare inequality propagates edge-level drift to an entire orbit at a cost proportional to $1/\eta$. Cyclic audits can additionally yield zero pair-level leave-one-out error without holding out any semantic object. To address these failures, we derive exact block-Woodbury leave-one-orbit-out updates and introduce a source-disjoint deployment gate over finitely many candidate readers. The gate retains the original reader unless uncertainty bounds certify lower drift within a prescribed clean-utility budget. etc..

---

248. Semantic Fibers and Cross-Gram Interference: A Calculus of Safety Drift in Overcomplete Representations

• Authors: Mohammed Ahnouch , Lotfi Elaachack
• URL: https://arxiv.org/abs/2609.14861
• Abstract:

  A deployed language model may refuse a harmful request in English yet comply with its faithful translation, revealing a cross-lingual safety failure that cannot be characterized reliably by output behavior alone. We formalize this phenomenon through an audited equivalence relation and show that, for a declared quotient, representation, metric, feature dictionary, scoring head, threshold, and contrast model, the resulting safety drift admits an exact linear-algebraic characterization. Specifically, the drift is a cross-Gram functional of the within-fiber contrast; its worst admissible value is a support function, while margin invariance is characterized by an annihilator condition. We introduce an intrinsic calibrated exposure measure, governed by the leverage duality $\chi^2=1/\ell-1$, which separates observed drift into three diagnostically distinct regimes: a reader fault removable by recalibration, an exact correction that is too ill-conditioned to be reliable, and a representation-level collision that no readout-only intervention can remove. Thus, identical observed exposure can lead to fundamentally different remediation verdicts. The framework also extends to cone-valued safety heads. An untied order-swap identity provides a diagnostic for the linear control interface; its calibration-state residual predicts a distinct three-control composition error on unseen states and targets, achieving median Spearman correlation $0.964$, compared with $0.269$ for a static cross-Gram baseline. etc…..

---

249. One Example Is Enough to Pass Fairness Benchmarks: Rethinking Fairness Evaluation for Aligned LLMs

• Authors: Naihao Deng , Samee Arif , Shuaichen Chang , Yulong Chen , Rada Mihalcea
• URL: https://arxiv.org/abs/2609.14860
• Abstract:

  Warning: This submission studies stereotypes and biases, and contains toxic and offensive examples, used for illustration purposes only. Fairness benchmarks such as BBQ have become the de facto standard for fairness evaluation across major model families. We argue that these benchmarks are too easy to support their role: training Qwen 2.5 7B Base with Group Relative Policy Optimization (GRPO) on a single BBQ example, or placing that example in context as a one-shot demonstration for in-context learning (ICL), lifts mean BBQ accuracy from 79.9% to 92.9% and 99.0%, respectively, closing 80% of the gap to its large-scale RLHF counterpart (96.1%) with GRPO, and surpassing it with ICL. These effects generalize across model families. A cross-conditioning analysis shows the improvement is carried by the reasoning traces generated by the model, and one example suffices to elicit a category-agnostic ``missing evidence’’ reasoning pattern. We argue that BBQ-style multiple-choice abstention benchmarks measure a single structural cue, and a model that solves them does not thereby become fair. We call for evaluation suites that cover a broader spectrum of fairness alignment.

---

250. RAIN: Region-Aware Inversion Network for Semantic Watermark Extraction

• Authors: Zilai Li
• URL: https://arxiv.org/abs/2609.14856
• Abstract:

  Semantic watermarks for diffusion models embed ownership information into the generative process while preserving perceptual quality, but Gaussian-Shading extraction conventionally requires multi-step diffusion inversion to recover the initial noise. Recent one-step methods show that this cost can be reduced substantially. We study this problem through extended flow matching and conditional regression. The key observation is that, near the high-SNR image endpoint, recovering a useful noise statistic given by the first-step output of the extended flow matching in the high-SNR regime is much simpler than reconstructing the full inverse trajectory, and Gaussian Shading only requires the recovered latent to remain in the correct watermark decision region. Based on this observation, we propose a lightweight, prompt-free extractor that decomposes endpoint recovery into an image-like anchor and a noise-oriented residual, which increases the capability of the model to utilize GPU parallel computation. The resulting method avoids iterative inversion and repeated evaluation of a diffusion-scale U-Net, providing an efficient one-step extraction pipeline with a concise theoretical interpretation. The computational cost of extracting noise is lower than that of both OSI and FARI. The github repo is there: this https URL

---

251. LLMs as Oracles: Reliance on LLMs for Subjective Personal Questions

• Authors: Myra Cheng , Lujain Ibrahim , Grace Liu , Michelle S. Lam , Vishakh Padmakumar , Nick Madibekov , Diyi Yang , Dan Jurafsky
• URL: https://arxiv.org/abs/2609.14849
• Abstract:

  We characterize how people are turning to LLMs as oracles: all-knowing authorities on subjective personal questions. Motivated by risks to users’ autonomy and well-being, we develop a typology and LLM-based methods to measure this form of AI reliance at scale and understand how people are offloading judgment and decision-making to AI. Applying our typology to public usage data (68K prompts from WildChat and ThoughtTrace), we find that LLM-as-oracle use has increased over time (2023-2026) and is more prevalent among younger users. We further build a privacy-preserving data donation tool to analyze individuals’ longitudinal usage data (140K prompts from 52 participants), identifying similar trends. People are often unaware of their own LLM-as-oracle use, and express dissatisfaction with this behavior after seeing our tool’s analysis. Finally, we identify two drivers of LLM-as-oracle use: people’s perceptions of AI and the behavior of AI models themselves, which motivate possible interventions to support users’ self-deliberation.

---

252. A Responsive Present, a Shared Past, a Social Other: Teens’ Overreliance on Companion AI Chatbots

• Authors: Mohammad Namvarpour (Matt), Tyler Chang , Afsaneh Razi
• URL: https://arxiv.org/abs/2609.14843
• Abstract:

  AI companions provide socially engaging interaction through availability, personalization, memory, roleplay, and emotionally responsive language. For teens, these systems may support sensitive self-disclosure, identity exploration, and relationship rehearsal while shaping intimacy expectations, offline relationships, emotional wellbeing, and self-understanding. We analyzed 17,053 verified quotations from 3,930 teen-relevant Reddit posts using thematic analysis. We identified 53 topics across seven thematic groups. Users described AI companions as sources of comfort, recognition, identity exploration, and relationship rehearsal, but also reported problematic attachment, social substitution, emotional dependence, and disruption to academic and social life. Roleplay, memory, perceived reciprocity, unwanted romantic or sexual role drift, privacy concerns, platform changes, and service interruptions shaped users’ boundaries and control. Awareness that the AI was artificial did not prevent guilt, obligation, grief, or distress. These findings show that companion-AI safety must address relationships over time through user-controlled memory, privacy, relational boundaries, and healthy disengagement.

---

253. Efficiency Hallucination: Formalizing and Measuring Behavioral Calibration in LLM-Based Code Optimization

• Authors: Sarah Wilson , Gail Kaiser , Patrick Musau
• URL: https://arxiv.org/abs/2609.14839
• Abstract:

  The integration of Large Language Models (LLMs) into automated code optimization introduces a critical reliability risk we term the Efficiency Hallucination: an LLM’s tendency to issue non-functional mutations with unsubstantiated performance claims on already-optimized code. This is driven by the Evaluation Trap, wherein binary benchmarks incentivize unnecessary modifications over safely abstaining. We present a validation framework using classification penalty methods, evaluated across 180 optimization runs on nine models (GPT, Claude, Gemini) using EffiBench. Under standard prompts, models exhibit a 100% over-edit rate on optimal code. Our guardrail raises correct abstention from 0% to to 44.4%, preserving a 100% edit rate on sub-optimal code with zero false abstentions. Calibration is uneven: GPT-5.4 Mini approaches near-perfect abstention, and simple code is recognized more reliably than complex code. Our framework offers a training-free mechanism to mitigate LLM overconfidence before deployment in production.

---

254. Enemray: Toward Capable Language Models for Hassaniya

• Authors: Cheikh Ahmed
• URL: https://arxiv.org/abs/2609.14829
• Abstract:

  We introduce Enemray, a Hassaniya-centric language model that enables general-purpose interaction in Hassaniya. Enemray is trained around a stability–plasticity objective: acquire strong Hassaniya linguistic and cultural competence while preserving the general reasoning, multilingual, instruction-following, and safety behaviors of a capable instruction-tuned model. The development pipeline separates language acquisition from behavioral specialization. A separately assembled continual-pretraining corpus provides broad exposure to natural Hassaniya and Mauritanian text; layer-selective continual pretraining learns a compact language-specific parameter update; that update is transferred into the instruction-tuned parameter space; and supervised post-training develops conversational, cultural, literary, task-oriented, and cross-lingual behavior. The supervised corpus integrates selected public Hassaniya and Mauritanian resources with a substantially larger body of newly collected, reconstructed, curated, and constructed instruction data, while policy-generated replay provides a retention signal from the reference model’s own behavior distribution. The resulting collection is substantially larger and broader in purpose than existing Hassaniya text resources. In evaluation, Enemray achieves the strongest English to Hassaniya translation among the compared open and proprietary models and the highest overall score on Mauritanian translation error detection, while retaining most of the general capabilities of its instruction-tuned base model on mathematical reasoning, knowledge, code generation, and function calling. This report describes the motivation, data construction, model design, training methodology, and evaluation of Enemray.

---

255. Route, Don’t Fix: Regime-Dependent Decoding Correction and a Trajectory-Gated Router for Reliable Clinical LLM Answer Selection

• Authors: Zeyu Dong , Benjamin Wang , Joyee W. Jin
• URL: https://arxiv.org/abs/2609.14825
• Abstract:

  Large language models (LLMs) are often deemed unsafe for clinical question answering because of their tendency to hallucinate. Retrieval augmentation, fine-tuning, and external verifiers require new infrastructure that clinical governance must approve and may add latency or extra model calls. Inference-time correction uses the model’s internal logit signals, but a fixed transformation need not suit every question. A corrector that improves accuracy by about ten percentage points on a truthfulness stress test yields negligible gains on clinical multiple-choice benchmarks, where instruction tuning concentrates output probability on one answer and leaves low terminal entropy. We introduce ALTAS, which reads terminal entropy and late-layer linearity ($R^2$) from one forward pass to choose per question between greedy decoding and late-layer trajectory correction. No classifier, probe, or head is trained; the router operates on candidate-answer logits and adds 6.5% latency overhead. Applied to every question, the correction improves TruthfulQA over greedy at 3B and 8B by 11.4 and 10.0 percentage points, respectively ($p<10^{-10}$). Gated per question, ALTAS retains gains of 8.3 to 9.5 percentage points while keeping MedQA, PubMedQA, and MedHallu within a one-percentage-point do-no-harm band, with no statistically significant differences from greedy. The method passes verification sweeps over frozen thresholds, the scoring rule, and the domain label.

---

256. A primer on evaluation methods for large language models in healthcare

• Authors: Suzannah E McKinney , Phuc Vu , Samuel A Justice , Christopher Humphries , Alyssa Pradhan , Timothy J Keyes , Bernardo C Bizzo , Keith J Dreyer , Sarah F Mercaldo , James M Hillis
• URL: https://arxiv.org/abs/2609.14819
• Abstract:

  Large language models (LLMs) have a growing range of applications in medicine, and their evaluation is critical for ensuring they provide benefit and not harm. This evaluation can be more challenging than traditional machine learning for many reasons, including probabilistic and open-ended outputs, and behavior that shifts with prompt design and accumulated context. This review covers four key areas of LLM evaluation: principles of study design, statistical methods, capability evaluation and clinical context evaluation. Capability evaluation considers different benchmarks, including multiple-choice, agentic and multi-turn benchmarks, alongside operational metrics like token usage. Clinical context evaluation addresses establishing accuracy of free text outputs, such as human review and LLM-as-a-judge, and clinical trial approaches. Across sections, we describe underlying concepts and potential pitfalls, while emphasizing the importance of aligning evaluation methods with the research question. Together, this article aims to provide a pragmatic basis for designing and executing rigorous evaluations of healthcare LLMs.

---

257. Mind Which Bird You Favour: Parameterizing Adequacy-Fluency Balance in Meta-Evaluation of Machine Translation

• Authors: Behzad Shayegh , Niloofar Kazemi
• URL: https://arxiv.org/abs/2609.14795
• Abstract:

  There is a tradeoff in machine translation meta-evaluation between prioritizing alignment with adequacy versus fluency. The balance depends on the combination of translation systems in the meta-evaluation dataset. This system set is a small, filtered sample whose characteristics change heavily across years and language pairs; it does not represent the true system distribution. Consequently, the adequacy-fluency balance is often unrepresentative and subject to change. For sensitive domains, controlling this balance is critical. We expose this balance as a tunable choice. To achieve a target balance, we reweight existing systems while minimizing distortion from uniform weighting, ensuring the evaluated systems remain real and representative. We provide an exact optimization algorithm with theoretical guarantees and pruning mechanisms to compute these weights. To validate meta-evaluation internal consistency, we design a scorer-augmentation framework that establishes a known relative identity for the scorers. Results demonstrate that our reweighting method effectively controls the adequacy-fluency balance and preserves the internal consistency, outperforming prior approaches. Finally, we analyze the performance of popular scorers across a sweep of this parameter.

---

258. The Stochastic Deputy: Structural Tenant Isolation for Tool-Using LLM Agents

• Authors: Mirza Samad Ahmed Baig , Syeda Anshrah Gillani , Asher Ali , Muhammad Hamzah Siddiqui
• URL: https://arxiv.org/abs/2609.14780
• Abstract:

  Multi-tenant tools commonly accept a tenant identifier and validate it against the caller’s entitlement. For a large language model (LLM) agent, that pattern delegates resource selection to a process whose context may contain attacker controlled instructions. We formalize this stochastic deputy problem and present a structural defense: remove tenant identity from the Model Context Protocol (MCP) tool schema, bind scope to a verified credential, and enforce it below the agent. In a 373-trial ablation across eight model configurations and two transports, a correctly validated tenant parameter served every out-of-scope attempt: 26 of 26, or 26 of 41 plausible-pretext trials overall. With the parameter removed, no tool signature could express the read. Twelve of 56 trials instead escaped the interface by forging writable scope, showing that interface invariance requires cryptographically protected context. On a production dataset containing multiple GBs of data, set-valued scope caused a measured $57\times$ latency ratio under function-wrapped membership predicates; a JSON_TABLE lateral join recovered index access where the tenant key was indexed. The evaluation also exposes deployment limits, including an entitlement-size query-planner cliff and incomplete index coverage. The result is a tenant-isolation argument that depends on enforceable interfaces and credentials rather than model compliance.

---

259. How broad is that claim? Mapping Generalisation in NLP Research

• Authors: Chenxin Diao , Nataliya Stepanova , Emily Allaway
• URL: https://arxiv.org/abs/2609.14770
• Abstract:

  Generalisations are common in scientific communication, even though they are semantically ambiguous. An automated method is needed to identify and categorise claims according to their level of generalisation, in order help detect an over-reliance on generalisations and possible misrepresentations of scientific findings. We introduce a comprehensive taxonomy of generalisations in the scientific domain, NLPGenX, which labels claims according to their level of generality and framing within the text. We operationalise this taxonomy with an LLM-powered framework, NLPGenA, that automatically classifies sentences from scientific articles into 5 different generalisation classes. We validate our framework with human annotators and use the framework to construct a large-scale dataset of NLP papers annotated according to generality, with auxiliary labels for hedging and vague descriptors (NLPGens). We use NLPGens to analyse the use of generalisations in NLP papers across multiple venues and subdomains, and to examine associations with citation counts, hedging, and vague descriptors.

---

260. Loop-Back Authority in LLM Agent Teams: A Paired Experiment on Flat and Hierarchical Coordination

• Authors: Burak Agachan , Max van Duijn , Amirhossein Zohrehvand
• URL: https://arxiv.org/abs/2609.14767
• Abstract:

  Hierarchical orchestration, in which a Manager agent reviews worker output and can send it back for revision, is the default coordination pattern in production multi-agent LLM frameworks. Classical organizational theory predicts that the authority link speeds convergence on decisive output; work on sycophancy and Degeneration-of-Thought predicts that authoritative critique makes LLM output worse. Prior comparisons vary whole frameworks on tasks with checkable answers, leaving the authority link untested on open-ended work. We present a paired experiment that holds five LLM agents, their roles, prompts, tools, models, and data fixed and varies one link: whether the Manager may reject a worker’s output and oblige a revision. Across 43 paired products and 86 runs of a business-intelligence reporting task, a five-model judge panel and a deterministic specification check score every report. The flat organization scores higher on Utility (d = 0.42, p = 0.009) and on Writing Clarity (d = 0.34, p = 0.030); the classical prediction fails. The reports are the same length, but hierarchical reports hedge 53% more, each revision loop is associated with a 0.14-point drop in Writing Clarity, and the hierarchical Writer’s first draft is indistinguishable from the flat report: the gap opens inside the revision loop. Specification accuracy is at ceiling in both organizations, and the supervisory tier costs 51.5% more tokens for no quality gain. A supervisor pays for itself when it can verify and becomes a liability when it can only opine.

---

261. TriCalRAG: A Three-Strategy, Retrieval-Augmented Benchmark for On-Premise LLM-Based Root Cause Analysis in AIOps

• Authors: Rohit Patel , Susil Kumar Mohanty , Jeenal Chaudhary
• URL: https://arxiv.org/abs/2609.14762
• Abstract:

  Cloud-hosted large language models (LLMs) are increasingly used for root cause analysis (RCA) in AIOps pipelines, but they introduce data privacy risk, network latency, and per-query cost that scale poorly with production log volumes. We present TriCalRAG, a benchmark evaluating open-weight LLMs served locally via vLLM on a single high-memory workstation GPU (NVIDIA RTX PRO 6000, 96GB) against a classical LSTM-based log anomaly detector (DeepLog), across four real, publicly available log datasets (BGL, HDFS, Thunderbird, OpenStack). We evaluate two open-weight models (Qwen2.5-14B, Mistral-Small) under three prompting strategies: zero-shot, few-shot, and retrieval-augmented generation (RAG) over a labeled incident history, reporting accuracy, precision/recall, and F1 with bootstrap 95% confidence intervals across 3 random seeds, alongside throughput and VRAM footprint. Our results show that RAG not only improves mean F1 by 0.10-0.27 over zero-shot prompting but, more importantly, substantially stabilizes model calibration: zero-shot prompting drives both models toward near-degenerate behavior (predicting “anomaly” on up to 100% of incidents on some datasets), while RAG keeps predicted-positive rates close to the true class balance in the majority of configurations. Mistral-Small achieves higher macro-averaged F1 than Qwen2.5-14B (0.644 vs. 0.560) but exhibits calibration failures in more configurations (7 vs. 5 of 12), while running at roughly half the throughput - indicating the better model choice depends on whether a deployment prioritizes peak accuracy or predictable behavior across prompting conditions. Ablations show batching scales throughput 41 times on a single card and that 4-bit quantization reduces latency 20% with no measurable accuracy loss. We release our benchmark harness, dataset splits, and evaluation code to support reproducible on-premise AIOps research.

---

262. From Visual Feedback to Textual Reviews: A Multi-Agent Vision-Language Framework for Image-Grounded Review Assistance

• Authors: Utsav Kumar Nareti , Ayush Bansal , Kumari Priya , Chandranath Adak , Soumi Chattopadhyay , Muhammad Saqib , Saeed Anwar
• URL: https://arxiv.org/abs/2609.14761
• Abstract:

  Visual feedback in the form of user-uploaded images and videos is becoming increasingly common in e-commerce platforms because it provides authentic evidence of product quality, defects, packaging conditions, and real-world usage. However, visual feedback alone often lacks the contextual explanations and subjective opinions necessary for informed decision-making, while many users provide limited textual feedback due to the effort required to compose detailed reviews. To bridge this gap, we introduce image-grounded review assistance, a novel task that aims to generate editable review drafts from user-uploaded product images. Unlike conventional image captioning, which focuses on objective visual description, the proposed task requires product-specific understanding, sentiment estimation, and evidence-driven review composition under challenging real-world conditions, including degraded image quality, excessive zoom-in, target ambiguity, and partial product visibility. We propose a multi-agent vision-language framework consisting of four specialised roles: product grounding, visual sentiment estimation, visual evidence generation, and review synthesis. The framework employs explicit intermediate representations, including product entities, predicted ratings, and evidence summaries, to improve interpretability and visual grounding. Experiments on a curated subset of the Amazon Reviews Electronics dataset demonstrate the feasibility of generating coherent, product-aware, and sentiment-aware review drafts from visual feedback. To the best of our knowledge, this is the first study to formulate image-grounded review assistance as a multi-agent vision-language reasoning problem, providing a practical step toward AI-assisted review authoring in e-commerce systems.

---

263. Refusal Reads Only a Slice of What the Model Knows: Harm-Keyed Routing and Its Exceptions Across Model Families

• Authors: Orion Reblitz-Richardson
• URL: https://arxiv.org/abs/2609.14759
• Abstract:

  Alignment applied after pretraining is shallow in a measurable way: a single direction in a model’s residual stream can be edited out, and the model stops refusing harmful requests. That fact says how easily refusal can be removed, not what the refusal decision was reading in the first place. We ask what it reads, and we separate that from what the model comprehends. Across four open-weight models spanning three families, moral comprehension is native to pretraining: a low-rank moral subspace crystallizes during pretraining, and alignment rotates it once without rebuilding it. The refusal gate, in contrast, is a fresh post-training construction with only a weak pretraining precursor, written into a narrow control-token channel where the refusal decision is orthogonal to the moral-judgment decision. The central result is causal and comes from one model, OLMo-3. A nested interchange rank sweep patches successively larger slices of the moral subspace between matched requests and reads how much of refusal’s response transfers: as the basis widens, moral judgment keeps reading more of it, while refusal levels off at the level of a single harm direction, and about three-quarters of refusal’s causal input lies outside the moral subspace altogether. Refusal reads the harm percept, not the moral content that judgment reads on the same patches. The picture is not uniform across families. Llama reads broad moral content; Qwen reads beyond the single harm cue but is unresolved at our sample size; GPT-OSS reads harm, and its refusals can be argued in either direction by its own reasoning trace. Where refusal reads only a low-rank slice and routes around the bulk of what the model knows, a rank-one edit removes it. Whether widening what refusal reads would also deepen the behavior is the open question this raises.

---

264. Calibrating Interpretability Instruments Before Trusting Their Verdicts

• Authors: Orion Reblitz-Richardson
• URL: https://arxiv.org/abs/2609.14754
• Abstract:

  Causal claims about large language model (LLM) internals rest on measurements. Those might include a projection, a cosine, an ablation delta, or an interchange patch among others. These measurements fail in specific, diagnosable ways that return a plausible number instead of an error, so a broken instrument can easily read as a finding. A covariance-matched null can saturate until every direction looks typical, a per-head attribution can overshoot the true residual write threefold on reordered-normalization architectures, an interchange patch can go sign-chaotic because its outcome is pinned at a ceiling, or a read-from verdict can be an artifact of measuring past the layer where the model already decided. This note documents six such failures from a causal interpretability program on refusal and moral representation, spanning several papers and a four-model open-weight panel; each mode is established on one or two of the four. For each we give the tell that catches it and a protocol keyed to a detectable trigger (reordered normalization, massive activations, a low-dimensional decision channel), so we and readers can check whether a given setup is exposed. The discipline reduces to four moves: calibrate against a positive-control ladder, certify with an orthogonal cell, compute power before spending compute, and state every read-from verdict at a depth referenced to the model’s commitment. The evidence is four architectures across three families within a single program; external replication across programs is future work.

---

265. A property-registry contract for retrieve-or-refuse thermal-mechanical lattice search

• Authors: Shaoliang Yang , Henry Chu , Zu Yashengjiang , Jun Wang
• URL: https://arxiv.org/abs/2609.14741
• Abstract:

  Early thermal-mechanical lattice requirements are knowledge-intensive and often jointly unsatisfiable: an engineer asks for a cell that is light, stiff, laterally conducting and cheap, and no cell in the library satisfies it. A design system should say so, and say which requirement to loosen and by how much, rather than return the nearest row. A generative model can return a candidate even when the library holds none. This work applies established conflict diagnosis to a catalogue of homogenised properties. Search over 1,397 homogenised unit cells and 19 base materials (26,543 combinations) returns a catalogue row that a second solver can rebuild, or, for an empty feasible set of at most eight constraints, inclusion-minimal unsatisfiable subsets (MUS) and the slack of a repair, not a neighbour. One registry declaration generates the prompt and the evaluator, so undeclared parsed keys cannot reach search. On a frozen suite of 64 typed queries, retrieve-or-refuse matches min-repair on every feasible query (48/48) and refuses every empty one (16/16) with MUS and slack. Constraint-ignoring nearest-neighbour and penalty search violate stated cost or density limits. On 216 empty queries, repairs printed at three significant figures and rounded outward stay feasible in every case (216/216). When density and cost must be kept, a list of minimum-cardinality repairs keeps them on 209 of 211 queries; a protection-first repair keeps them on all 211, the same as full diagnosis. A 308-request parse benchmark is a template-text check, not unconstrained engineer prose.

---

266. Building Legal Reward Models for Grounding and Abstention

• Authors: Rilton Franzone , Valentin Noël , Puyu Wang , Philip Torr , Fabio J. Fehr
• URL: https://arxiv.org/abs/2609.14739
• Abstract:

  Large language models are increasingly used in high-stakes domains such as law, where systems must ground their reasoning in retrieved evidence and abstain when that evidence is insufficient. However, existing reward models are largely optimised for general preferences rather than contextual grounding, limiting their ability to evaluate these behaviours in retrieval-augmented generation (RAG) settings. We introduce a framework for transforming existing legal QA datasets into contextual preference data and use it to construct LegalRewardBench (LRB), a benchmark for evaluating grounded legal generation under noisy and insufficient retrieval conditions. Across general and legal contextual evaluation, we find that contextual DPO improves grounded evaluation, but performance is sensitive to preference-data construction. Length-balanced augmentation substantially improves grounded legal evaluation, with the strongest configuration combining length-balanced legal and general contextual preference data and improving performance by up to $\mathbf{+25.6}$pp over baseline. We further find evidence of cross-jurisdiction transfer: models contextually refined primarily on Victorian criminal-law data improve grounded evaluation on external US legal benchmarks, including a $\mathbf{+16.2}$pp improvement on \textsc{Housing Statute QA}. Together, these results provide a reproducible foundation for constructing and evaluating grounded legal reward models in retrieval-augmented settings.

---

267. OCT-FedSIR: Toward Trustworthy Federated Ophthalmic Learning under Annotation Noise

• Authors: Sina Gholami , Abdulmoneam Ali , Tania Haghighi , Rashadul H. Badhon , Behafarin Emam , Sally S. Y. Ong , Atalie C. Thompson , Theodore Leng , Ahmed Arafa , Jennifer I. Lim , Minhaj Nur Alam
• URL: https://arxiv.org/abs/2609.14734
• Abstract:

  Federated learning enables collaborative model development without centralizing patient data, but annotation reliability at participating institutions cannot always be assumed. In ophthalmic imaging, differences in disease prevalence and class composition can resemble changes caused by corrupted supervision. We introduce OCT-FedSIR, a reliability-aware spectral framework for federated OCT classification under client-dependent annotation noise and heterogeneous data distributions. OCT-FedSIR combines class-balanced spectral estimation, Stage-I logit adjustment, complementary spectral descriptors, selective spectral relabeling, and noise-aware federated optimization. We evaluated the framework on the Kermany, University of Illinois Chicago, and Wake Forest datasets under symmetric and structured asymmetric noise and three levels of non-IID heterogeneity. Across 117 experimental conditions, OCT-FedSIR achieved a mean accuracy of 86.73%, compared with 79.94% for RoFL and 78.75% for FedCorr. It correctly separated clients with original and corrupted annotations across all evaluated conditions, while the original FedSIR identification procedure was less robust, particularly under asymmetric noise. Spectral relabeling recovered 77.2% of corrupted annotations with 91.3% correction precision and a 3.5% false-correction rate. Retaining corrected clients outperformed spectral pruning by 9.30 percentage points on average. These findings show that annotation noise can often be identified and corrected without discarding informative client data.

---

268. Carryover Drafting: Recycling Rejected States for Speculative Decoding

• Authors: Jahyun Koo , Sunghyeon Woo , Jaeeun Kil , Jeongtae Lee , Sungjae Lee , Kyomin Jung , Minsub Kim
• URL: https://arxiv.org/abs/2609.14717
• Abstract:

  Speculative decoding accelerates LLM inference by verifying multiple drafted tokens in parallel, allowing a single target forward pass to accept several tokens. By construction, verification computes representations for both accepted and rejected tokens. Yet, conventional drafters retain only the representations of accepted tokens, leaving the substantial verifier computation spent on rejected tokens effectively wasted. We find that these discarded hidden states generated during target forward retain useful information about future tokens that can improve subsequent drafts. However, realizing this opportunity poses two distinct challenges. At inference, recycling overhead can increase drafting latency, diminishing the speedup gained from increased acceptance length. During training, standard parallel drafter training does not produce inference-aligned rejected states, while obtaining them through sequential rollouts would sacrifice parallelism across training positions. We introduce Carryover Drafting, which addresses both challenges. Carryover recycles rejected target hidden states as temporary KV context, allowing the drafter to selectively attend to them. It reuses the drafter’s existing interface and adds only a single learned embedding to distinguish rejected states from committed context. The additional KV context is replaced each drafting round, keeping its length bounded by one proposal block. We introduce parallel draft–verify–draft training that exposes the drafter to inference-aligned rejected states while preserving parallelism across training positions. Experiments with DFlash and a DSpark-derived semi-autoregressive drafter across two target models show that this simple Carryover mechanism improves average acceptance length by 6.5–14.7% and end-to-end vLLM speedup by 7.9–14.4% over the corresponding baselines, with speedup gains reaching 28.8% on translation.

---

269. WaterKron and FlipFlop Hessian: Information-Theoretically Grounded Quantization with Kronecker-factored Hessians

• Authors: Johann Birnick , Rayan Saab
• URL: https://arxiv.org/abs/2609.14706
• Abstract:

  How should a Kronecker-factored Hessian approximation be chosen for post-training quantization? We address this question through WaterKron, which combines two-sided GPTQ with row- and column-dependent waterfilling scales and entropy coding. We derive its high-rate distortion with respect to the full Hessian using an explicit Kronecker-Hessian mismatch factor $\Phi$. This factor quantifies the asymptotic distortion penalty due to the Kronecker Hessian approximation and provides a criterion for selecting the factors optimally. Minimizing $\Phi$ leads to a Gaussian covariance-fitting problem with classical ``flip-flop’’ updates. We thus give a rate-distortion justification for using the resulting FlipFlop Hessian in quantization. We evaluate it empirically, finding that the FlipFlop Hessian consistently improves KL divergence and perplexity over input-only, marginal, and Frobenius-based Hessian choices.

---

270. PU classification under Non-SCAR: clustering-assisted logistic model with oversampling enhancement

• Authors: Konrad Furmańczyk , Kacper Paczutkowski
• URL: https://arxiv.org/abs/2609.14675
• Abstract:

  This study addresses the PU classification problem under violations of the SCAR assumption. We investigate logistic regression-based approaches, namely the cluster method and its extensions with strict and non-strict Lasso regularization. The primary contribution of this work is the integration of the SMOTE technique to alleviate class imbalance and systematically assess its impact on the performance of the considered algorithms. SMOTE is first applied to rebalance the training dataset. Next, cleaning labels are derived via 2-means clustering. Logistic regression is then trained on the cleaned data, where identified positive instances are augmented with additional true positives and the remaining observations are treated as negative. The experimental evaluation is conducted on 13 real benchmark datasets and one synthetic dataset. For comparison, we include the naive approach and the Spy-EM method. The results demonstrate that incorporating SMOTE improves classification performance when the SCAR condition is violated and indicate moderate robustness of the LassoJoint method in this setting.

---

271. MANE: A Multi-Path Adaptive Network for Edge Onloading of Deep Neural Networks

• Authors: Sokratis Nikolaidis , Stylianos I. Venieris , Leonidas Malachias , Iakovos S. Venieris
• URL: https://arxiv.org/abs/2609.14660
• Abstract:

  Split computing constitutes a widely used distributed inference approach, where a lightweight head model is onloaded onto the device and a heavier tail model resides on an edge server, leveraging the growing computational capabilities of modern System-on-Chips while alleviating server load. As intelligent indoor environments such as smart offices grow increasingly populated with diverse IoT devices, a single edge server must simultaneously assist multiple devices, each competing for the same shared inference resources. Without a principled mechanism to manage this shared load, the server is quickly overwhelmed, causing latency SLO violations and rendering server-assisted inference ineffective. In this work, we present MANE, a distributed inference framework that equips the server with a multi-path tail architecture, enabling a dynamic accuracy–throughput trade-off at runtime. By introducing a novel multi-path model architecture, a three-stage training scheme featuring a Joint Head Network Distillation loss and a hysteresis-based scheduler with an equitable device-fallback policy, MANE maintains over 80% SLO satisfaction rate where state-of-the-art onloading methods fail completely, while preserving accuracy 6pp higher than on-device alternatives, across up to 40 concurrent devices.

---

272. Compositional SVG Generation via VLM-Driven Hierarchical Semantic Parsing

• Authors: Sehwan Park , Taehoon Kim , Geonhee Han , Dohyun Kim , Seung Wook Kim , Paul Hongsuck Seo
• URL: https://arxiv.org/abs/2609.14657
• Abstract:

  While Vision-Language Models (VLMs) excel at visual reasoning, generating structured, editable Scalable Vector Graphics (SVG) remains a fundamental challenge. Existing pipelines predominantly yield flat, semantically agnostic collections of paths, where editing a single object requires manually identifying its constituent paths. To address this, we propose a VLM-driven agentic framework for semantic compositional SVG generation. Our pipeline recursively parses visual scenes into semantic and geometric hierarchies via top-down decomposition, visual grounding, and prompt-driven amodal occlusion recovery, ensuring each component is geometrically complete. Furthermore, we introduce the Semantic SVG Benchmark with human-annotated semantic groups and novel sub-component metrics (Semantic Recall/Precision, PERE) to explicitly evaluate structural compositionality and functional editability. Experiments show that our natively predicted structures surpass the upper bounds of existing flat-generation methods in both grouping quality and editability, while maintaining state-of-the-art visual fidelity.

---

273. Natural Language Knowledge Graph Query Execution: Leveraging Controlled Semantics in the LLM Context Window

• Authors: Blake G. Fitch
• URL: https://arxiv.org/abs/2609.14652
• Abstract:

  Large Language Model (LLM) applications often transfer domain concepts into the model’s context informally, through prompt prose, schema dumps, and examples. We show that for database queries, data model concepts pass to LLMs more effectively through representations whose vocabulary terms carry declared, machine-readable semantics (controlled semantics). NLKGQ is a working system and reusable framework that does this for data modeled in a knowledge graph. A formal OWL ontology serves as the transfer mechanism, concentrating the meaning of the data into semantically precise tokens the model can use directly. In a single LLM call, NLKGQ places in the context a system prompt instructing on SPARQL, the complete domain OWL ontology, and a domain-specific prompt addition, together with the user’s natural language query. The model then generates the SPARQL query directly, zero-shot. Where the native vocabulary of an existing database or federation of databases is opaque, a wrapper ontology substitutes clean terms and a runtime rewriter restores the native forms. Evaluating on DBLP-QuAD 2.0 showed that its scores depend on the graph snapshot, the endpoint used, and the wording of its machine-generated questions, so we propose DBLP-QuAD 3.1, which maintains the intent of 2.0 while making reference results deterministic, revising reference SPARQL where needed, and rewriting the natural language questions, with a frontier model, to state each reference query’s intent clearly and completely. We evaluate on the DBLP-QuAD 2.0 benchmark (57.6% Match under deterministic re-scoring), DBLP-QuAD 3.1 (89.9% Match on 1,000 questions), SemOpenAlex (98% Match against a published baseline’s 86% on the identical test set), and neuroimaging metadata (100%).

---

274. Skill Composition for Legged Robot Reinforcement Learning

• Authors: Daniel Gigliotti , Flavio Maiorana , Fabio Patrizi , Luca Iocchi
• URL: https://arxiv.org/abs/2609.14647
• Abstract:

  Robots, and humanoid robots in particular, are increasingly competent at individual behaviors, each obtained by training a specialized controller. A specialized skill is quick to train, converges reliably because the problem it faces is narrow, and can be validated on its own, none of which is true of a single end-to-end policy asked to cover everything. What remains fragile is the transition between them. We argue that the composition of independent sub-policies deserves to be treated as a research problem in its own right, rather than as an implementation detail left to whatever mechanism happens to be at hand. Reliable composition is what turns a collection of separate skills into a repertoire that can be used, extended and shared. More fundamentally, if control can be passed between specialized policies safely, and at any moment, the choice of what the robot should do next can be delegated to a component of an entirely different nature, such as a planner, an automaton or a symbolic controller, whose behavior can be inspected in advance. The policies would then only ever have to act, and what the robot can be trusted to do would become verifiable.

---

275. CompCQR: Compositional Query Generation for Training-Free Conversational Search

• Authors: Yunah Jang , Kang-il Lee , Joongbo Shin , Kyomin Jung
• URL: https://arxiv.org/abs/2609.14646
• Abstract:

  Multi-turn interactions with LLMs are becoming increasingly common in information-seeking scenarios. However, user queries are often ambiguous and context-dependent, making them ill-suited for direct use as retriever queries. Conversational query reformulation (CQR) addresses this issue by rewriting the current utterance into a stand-alone query grounded in the dialogue history. Recent LLM-based CQR approaches achieve strong performance; however, their repeated LLM invocations and misalignment with downstream retrievers remain challenges. In this work, we begin from the observation that retrievers are highly sensitive to content ordering: simply reordering the same content can lead to changes in retrieval coverage and performance. Based on this, we propose a novel training-free method that generates a very large number of queries with minimal LLM usage by compositionally combining a small set of atomic components. We further apply LLM reasoning to construct a high-quality document set that balances precision and recall while capturing the user’s core intent. Our framework generalizes across both open- and closed-source LLMs as well as dense and sparse retrievers. It achieves strong performance on four widely used conversational benchmarks, with up to 22.5% relative MRR improvement over the previous state-of-the-art baseline with far fewer LLM calls.

---

276. Investigating the Impacts of Generative AI on Information Seeking

• Authors: Alexi Orchard , Shannon Lodoen
• URL: https://arxiv.org/abs/2609.14638
• Abstract:

  This paper is an encore submission of our 2026 journal article “Expertise and Information Seeking in the Age of Generative AI: New Procedures, New Problematics” with an extended discussion for the CSCW 2026 “Broader Impacts of GenAI in Communication” Workshop on October 10, 2026. In the original article, we employ procedural rhetoric to analyze how generative AI chatbots leverage natural language signifiers of expertise and intelligence to influence users’ perception of their trustworthiness. In this submission, we extend our conversation in the CSCW community with the goal of cultivating a cross-disciplinary vocabulary for describing, analyzing, and mitigating the risks posed by the integration of generative AI into human communication practices. It is important to develop an understanding of how the procedures surrounding information-seeking practices are informed by users’ values, experiences, and expectations - and how these procedures might in future be altered by the emerging turn toward AI “experts” and authority.

---

277. Open-UniMo: Towards Unified Motion-Language Understanding and Generation in the Open World

• Authors: Guocun Wang , Kenkun Liu , Guorui Song , Jing Lin , Zhe Huang , Luyuan Zhang , Dake Zhong , Choo Sin Wai , Xiaoguang Han , Haoqian Wang
• URL: https://arxiv.org/abs/2609.14615
• Abstract:

  Unified motion generation and understanding is crucial for embodied AI systems that can both synthesize and interpret human actions in open-world environments. Existing motion-language models often treat motion as an auxiliary modality of a language model, leading to text-dominated representations and limited cross-modal interaction. Moreover, the next-token prediction paradigm is not naturally suited to long motion sequences, where autoregressive generation may accumulate prediction errors. To address these challenges, we propose Open-UniMo, a unified Large Motion-Language Model (LMLM) trained on million-scale open-world motion-language data. Open-UniMo promotes modality parity by extending Qwen’s vocabulary of about 150K text tokens with 64K motion tokens, enabling motion and language to share a unified token space. We further introduce motion-consistent Chain-of-Thought reasoning as an intermediate representation to bridge language semantics and motion dynamics. Open-UniMo is trained with a two-stage pipeline, where supervised fine-tuning establishes CoT-guided bidirectional motion-language mapping and Group Relative Policy Optimization (GRPO) improves semantic alignment while mitigating cumulative errors in autoregressive motion-token generation. To support comprehensive evaluation, we propose Open-MoBench, a VLM-guided benchmark for assessing text-to-motion (T2M) generation, motion-to-text (M2T) understanding, and bidirectional consistency. Extensive experiments show that Open-UniMo achieves state-of-the-art performance on both conventional metrics and Open-MoBench. Furthermore, ablation studies reveal that M2T understanding is not primarily limited by motion-token vocabulary size; instead, coupling M2T with the learnable T2M generation path yields stronger cross-modal representations, demonstrating that generation can facilitate understanding in AR-based motion-language modeling.

---

278. Diagnosing Temporal Misalignment in Multichannel Time-Series Classification with Minimum Description Length

• Authors: Sebastian Buschjäger , Michael Frichert , Daniel Kuhe , Jian-Jia Chen
• URL: https://arxiv.org/abs/2609.14595
• Abstract:

  Multichannel time-series classification commonly assumes synchronized sensor streams, although latency, clock drift, and preprocessing can introduce relative delays during data collection or after deployment. Existing synchronization solutions are often hardware-specific and difficult to apply retrospectively. Consequently, synchronization problems may remain undetected while classification performance is suboptimal. We introduce a classifier- and label-free diagnostic based on minimum description length (MDL). Our method applies candidate temporal shifts to sensor groups and measures how efficiently one group can be encoded through a representation of the remaining channels. An increased codelength indicates that the shift destroys shared temporal structure, whereas the minimum identifies the alignment most strongly supported by the data. Unlike learned synchronization methods, the diagnostic requires neither retraining nor a trusted aligned reference and can therefore test both training and deployment data for misalignments. Experiments on two controlled synthetic tasks and nine real-world datasets show that the metric exposes alignment structure and can recover accuracy under induced deployment drift. A whole-dataset audit further identifies stable nonzero MDL optima in established benchmarks including FordChallenge, Opportunity, PAMAP2, and UCIActivity, revealing potential systematic offsets that conventional model evaluation does not expose. Our method thus provides a general-purpose tool for detecting, understanding, and correcting temporal misalignment throughout the time-series learning pipeline. Our code is available under this https URL .

---

279. SENTINEL: A Multi-Pathway Architecture for Detecting Living-Off-the-Land APT Attacks on Windows Command Lines

• Authors: Ahad Bin Islam Shoeb , Kamrul Hasan , Jamal Uddin Tanvin , Liang Hong , Imtiaz Ahmed , Md Arif Billah , Al Amin
• URL: https://arxiv.org/abs/2609.14593
• Abstract:

  Living-Off-the-Land (LOTL) is the dominant evasion technique of Advanced Persistent Threat (APT) actors, exploiting legitimate Windows utilities to conduct malicious operations without deploying custom malware and enabling state-sponsored campaigns to maintain persistent access within military and critical defense infrastructure for extended periods. Existing detection methods fail against obfuscated commands and multi-stage attack sequences, as demonstrated by the Volt Typhoon APT campaign, which maintained undetected access to U.S. critical infrastructure for over 18 months using exclusively signed Windows utilities. We present SENTINEL, a multi-pathway architecture integrating BERT-based semantic encoding, character-level CNN for obfuscation invariance, inter-command attention for multi-stage pattern recognition, and autoencoder-based anomaly scoring. Evaluated on a balanced Volt Typhoon benchmark derived from Microsoft and CISA threat intelligence advisories, SENTINEL achieves 92.0% accuracy on documented state-sponsored attack commands and 91.2% on obfuscated variants, compared to 74.0% and 72.0% for standalone BERT. Per-class analysis reveals that models achieving over 98% overall validation accuracy on imbalanced data exhibit only 44-58% malicious recall on balanced adversarial sets. Character-level processing contributes 5.6 percentage points of obfuscation invariance, and the 8.0 percentage point gap over augmentation-only baselines confirms structural architectural value beyond data-driven robustness alone.

---

280. AlgoRAG: Retrieval-Augmented Generation for Theoretical Computer Science Education – A Comprehensive Evaluation Framework for Algorithm Analysis and Complexity Theory

• Authors: Sushan Adhikari
• URL: https://arxiv.org/abs/2609.14572
• Abstract:

  Teaching abstract theoretical computer science (TCS) concepts such as algorithm analysis and complexity theory is challenging because students must handle formal proofs and asymptotic reasoning that conventional resources rarely explain in an adaptive, on-demand way. We present AlgoRAG, a specialized Retrieval-Augmented Generation (RAG) system that couples a large language model (LLM) with a curated, domain-specific knowledge base to address these challenges. The knowledge base integrates authoritative textbooks, 847 lecture slides, 312 practice problems with solutions, 156 worked proof templates, and 89 complexity worksheets. AlgoRAG incorporates domain-specific optimizations including mathematical entity recognition, notation-aware retrieval, and pedagogical re-ranking. We evaluate AlgoRAG on 179 curated exam-style questions spanning asymptotic analysis, recurrence relations, dynamic programming, graph algorithms, NP-completeness, sorting, and divide-and-conquer. The system achieves a 100% success rate with a mean response time of 38.0 seconds. While BLEU-4 scores are zero – a known limitation of n-gram matching on mathematical proofs where equivalent reasoning may use entirely different notation – AlgoRAG attains ROUGE-1 F1 of 0.0963, ROUGE-L F1 of 0.0683, and a pedagogical quality score of 0.7620, indicating that responses are well-structured and didactically sound even when surface wording diverges from reference answers. Performance is especially strong on NP-completeness (ROUGE-1 F1 = 0.1285, pedagogical quality = 0.7643) and graph algorithms (ROUGE-1 F1 = 0.1023, pedagogical quality = 0.8086). These results support the conclusion that RAG is an effective architecture for personalized theoretical-CS instruction, providing correct, context-rich explanations even for highly abstract topics.

---

281. Disentangling Topology and Diversity in Multi-Agent LLMs for Multilingual Low-Resource Emotion Detection

• Authors: Ulugbek Shernazarov , Charitha Ruwansiri Weerakon Basnayake , Abdelkhaleq El Jarjini , Noel Crespi , Praboda Rajapaksha
• URL: https://arxiv.org/abs/2609.14570
• Abstract:

  Multi-agent LLM systems combine multiple inference calls, but prior work often confounds how calls are connected with how they are diversified. We study these factors independently: inference topology and source of inter-agent diversity. In a controlled $2 \times 3$ matrix, we cross parallel aggregation and sequential refinement with stochastic sampling, role prompting, and learned QLoRA specialization, under a fixed three-call budget and output protocol within each backbone. Using Qwen2.5-14B-Instruct and Llama-3.1-8B-Instruct, we evaluate all six configurations on multilingual low-resource emotion detection across nine languages. Parallel learned specialization is strongest on Qwen at 52.83 Macro-F1 and reaches 52.94 on Llama. On Qwen it also exceeds same-backbone zero-shot, few-shot, CoT, and seven-call self-consistency baselines. The preferred topology depends on diversity source: sequential refinement helps stochastic and prompted settings, while the learned Width advantage shrinks from 2.83 points on Qwen to 0.17 on Llama. Depth-wise analysis suggests that later learned specialists can overwrite correct early predictions, although the aggregate effect is backbone-dependent. Overall, how agents are differentiated produces larger performance shifts than topology, which should be evaluated jointly with specialization.

---

282. Proving olympiad geometry theorems on a superconducting quantum processor

• Authors: Ning Wang , Zheng-Zhi Sun , Zhengyi Cui , Yiren Zou , Aosai Zhang , Fanhao Shen , Jiarun Zhong , Zehang Bao , Zitian Zhu , Han Wang , Jia-Nan Yang , Jiayuan Shen , Gongyu Liu , Yanzhe Wang , Yihang Han , Yiyang He , Jiahua Huang , Sailang Zhou , Xinrong Zhang , Yaozu Wu , Zixuan Song , Jinfeng Deng , Hang Dong , Qi Ye , Weikang Li , Si Jiang , Yixuan Ma , Shuangyue Geng , Zhide Lu , Chao Song , Hekang Li , Pengfei Zhang , Qiujiang Guo , H. Wang , Dong-Ling Deng
• URL: https://arxiv.org/abs/2609.14533
• Abstract:

  Automated theorem proving seeks to use computational systems to prove or disprove mathematical and logical statements [1, 2]. It underpins a wide range of applications, and enhancing theorem-proving capabilities remains a central objective in artificial intelligence [3]. Although recent neuro-symbolic systems have achieved remarkable progress [4-7], their operation is ultimately constrained by classical computational architectures. Quantum computing [8], by contrast, enables information encoding and coherent parallelism beyond classical limits [9-14], raising the possibility of accelerating structured symbolic deduction [15]. Here we report the experimental realization of automated geometry theorem proving on a fully programmable superconducting quantum processor. We develop two complementary quantum proving frameworks. The first implements Wu’s algebraic elimination method using quantum pseudo-division, with multivariate polynomials represented in superposition states, enabling quantum algebraic theorem proving. The second implements the full-angle method as backward symbolic reasoning through a hybrid quantum strategy-guided architecture, demonstrating a general route toward quantum symbolic proof search. As illustrative examples, we prove two theorems on a superconducting quantum processor: the perpendicularity of the diagonals of a square and a 1978 International Mathematical Olympiad geometry problem. Our results establish, at the experimental level, automated logical reasoning as a viable task for near-term quantum processors and provide a concrete pathway toward quantum-enhanced symbolic intelligence.

---

283. Sharing standardized image-derived data in computational pathology using DICOM

• Authors: Daniela P. Schacherer (1), Christopher P. Bridge (2, 3), David Clunie (4), Igor Octaviano (5), André Homeyer (1), Markus D. Herrmann (3, 6), Olivier Gevaert (7), Tabita Ghete (8), Markus Metzler (8), Henning Hoefener (1), Tahsin Kurc (9), Curtis Lisle (10), Kenneth Philbrick (11), Joel Saltz (9), Yuanning Zheng (7), Andrey Fedorov (3, 12) ((1) Fraunhofer Institute for Digital Medicine MEVIS, Bremen, Germany, (2) Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, USA, (3) Harvard Medical School, Boston, USA, (4) PixelMed Publishing, Bangor, USA, (5) Radical Imaging LLC, USA, (6) Massachusetts General Hospital, Boston, USA, (7) Stanford University, Stanford, USA, (8) University Hospital Erlangen, Erlangen, Germany, (9) Stony Brook University, Stony Brook, USA, (10) National Institute of Allergy and Infectious Diseases, Bethesda, USA, (11) Google Research, Mountain View, USA, (12) Brigham and Women’s Hospital, Boston, USA)
• URL: https://arxiv.org/abs/2609.14530
• Abstract:

  Development and evaluation of computational pathology methods require access to large and diverse datasets. Over the past decade, various initiatives invested significantly into collecting, centralizing, and sharing pathology imaging data. In contrast, sharing of image-derived data such as region-of-interest delineations or segmentation masks is less well developed. In this work, we describe our approach to encoding and sharing image-derived pathology data in a standardized manner within the National Cancer Institute (NCI) Imaging Data Commons (IDC), a platform that hosts and provides public access to de-identified radiology and pathology data. The IDC relies on the Digital Imaging and Communications in Medicine (DICOM) standard for data harmonization, yet the adoption of DICOM for pathology image-derived content has remained largely unexplored until now. Here, we present five representative datasets harmonized by conversion from their original representations into DICOM and shared publicly in the IDC. We demonstrate the benefits of this harmonization, describe contributions to critical open-source tooling, and discuss technical considerations relevant to broader adoption of DICOM for pathology image-derived data.

---

284. EdgeHAR: An Edge-Native Compact Sensor Foundation Model for Human Activity Recognition

• Authors: He Zhang , Siyu Yuan , Siyu Liu , Sizhen Bian , Bin Guo
• URL: https://arxiv.org/abs/2609.14498
• Abstract:

  Sensor-based human activity recognition (HAR) is fundamental to ubiquitous and wearable computing, yet existing foundation models are largely designed for cloud-scale deployment and struggle with real-world sensing shifts, including unseen users, devices, sampling rates, and sensor placements. We present \textbf{EdgeHAR}, an edge-native compact sensor foundation model designed for wearable intelligence. Unlike conventional models that entangle activity knowledge with acquisition variations, EdgeHAR learns transferable representations by factorizing sensor signals into three latent codes: an \textbf{(i)Activity-Semantic Code} capturing reusable activity knowledge, a \textbf{(ii)Motion-Dynamics Code} modeling temporal patterns, and an \textbf{(iii)Acquisition-Context Code} representing sensor-specific variations. This disentangled design enables efficient adaptation to new users, devices, placements, and activity classes with limited target-domain data. By incorporating lightweight adaptation modules, EdgeHAR achieves foundation-model-level transferability while satisfying edge constraints in computation, memory, latency, and privacy. Experiments across heterogeneous HAR datasets demonstrate that EdgeHAR maintains competitive recognition performance under distribution shifts with substantially reduced deployment cost. EdgeHAR establishes a practical paradigm for compact, edge-first foundation models for ubiquitous sensing systems.

---

285. Bridging the Modality Gap in Long-Form Clinical Audio: A Comparative Study of Lightweight and Heavyweight End-to-End SOAP Generation

• Authors: Ziyu Zhang , Mingchen Shao , Wenjie Tian , Tianlun Zuo , Longhao Li , Lei Xie
• URL: https://arxiv.org/abs/2609.14467
• Abstract:

  Automating clinical documentation from long-form doctor-patient conversations remains challenging for modern audio-language models. While cascaded ASR systems perform well, end-to-end (E2E) models often struggle with information loss and hallucinations on extended audio. For the BeTraC 2026 challenge, the ASLP team presents a fully E2E multimodal system that generates structured SOAP notes directly from audio, bypassing intermediate transcripts. We constructed a 1.41-million-sample multi-task corpus and applied a multi-stage pipeline: domain pre-training, supervised fine-tuning, and reward optimization. Evaluating the architecture under both Lightweight (3B) and Heavyweight (30B) constraints reveals that each training stage progressively enhances performance. Furthermore, scaling to 30B parameters substantially boosts concept extraction and summarization quality. Ultimately, our E2E systems consistently outperform representative cascaded ASR+LLM baselines, proving the efficacy of direct multimodal optimization for clinical documentation.

---

286. Follow the Geometry, Not the Model: Cold Start Semi-Supervised Learning

• Authors: Itai David , Daphna Weinshall
• URL: https://arxiv.org/abs/2609.14451
• Abstract:

  Modern semi-supervised learning (SSL) couples pseudo-label generation and classifier training, using the classifier’s own confidence to select the pseudo-labels that are then used to update the model. In the cold-start regime, where at most a few labels per class are available, this coupling is ill-posed, since the classifier cannot supervise itself before it has learned. To address this problem, we propose VAST (Veracity-Aware Semi-Supervised Training), which decouples these two stages. Probabilistic beliefs over the unlabeled set are first inferred directly from the geometry of a frozen self-supervised embedding and only then distilled into an inductive classifier. The construction rests on the Veracity Matrix, a kernel-based structure that aggregates label evidence across the data manifold and admits an interpretation as a Dirichlet posterior under a per-observation powered-likelihood model. Additionally, we introduce Veracity Propagation, a self-terminating belief-spreading step that extends coverage beyond the kernel neighborhood of the labeled set. Under a controlled protocol in which all methods receive identical frozen embeddings and labeled sets, VAST outperforms the strongest graph-based SSL baselines at every operating point across three datasets, with statistically significant gains in 7 of 9 comparisons, while producing a deployable inductive classifier rather than requiring transductive graph inference. Compared with end-to-end confidence-gated SSL, we further find that these methods underperform in this setting and, in our experiments, do not consistently exceed labeled-only performance.

---

287. NeuroActiSep: Detecting Factual Hallucinations from Feed-Forward Neurons in a Single Pass

• Authors: Ali Derogar Odolou , Reza Nazari , Mostafa Salehi
• URL: https://arxiv.org/abs/2609.14448
• Abstract:

  Hallucination in large language models reduces their reliability and slows adoption. Various white-box studies have used internal representations to detect patterns of truthfulness and factuality. A less-studied approach is to identify feed-forward neurons correlated with hallucination. We propose a method to rank feed-forward neurons at the final prompt token using a custom neuron selection dataset. We transfer the selected neuron identities to train hallucination classifiers on other factual question answering datasets. Our work provides empirical evidence that probes trained using the features from the selected neurons perform on par with probes trained on internal states. We also analyze the distribution of selected neurons and the effect of layer depth on detection performance.

---

288. From Visual Attribution to Clinical Reasoning: Explainable Parkinson’s Disease Screening from Hand-Drawn Patterns

• Authors: Aritra Dey , Utsav Kumar Nareti , Chandranath Adak , Soumi Chattopadhyay , Krishna Gopal Sasmal , Saeed Anwar
• URL: https://arxiv.org/abs/2609.14441
• Abstract:

  Parkinson’s disease (PD) manifests early neuromotor impairments that become observable in controlled hand-drawn patterns such as spirals and meanders, where tremor-induced oscillations, stroke irregularity, and curvature instability reflect underlying motor degradation. In this work, we present an explainable framework for PD screening from offline hand-drawn patterns that integrates discriminative visual modeling with clinically grounded reasoning. The predictive model captures distributed structural distortions and fine-grained texture variations. It is evaluated under subject-disjoint protocols to ensure reliable generalization. To move beyond black-box classification, we introduce a multi-stage explainability pipeline that combines visual attribution with structured symptom abstraction. Salient regions are identified using attention- and gradient-based localization, followed by extraction of clinically meaningful motor descriptors quantifying contour roughness, curvature irregularity, stroke variability, and tremor-frequency energy. These descriptors are subsequently translated into coherent clinical rationales through a language-based reasoning module, linking model evidence to established PD symptomatology. By bridging visual attribution and clinical interpretation, the proposed framework advances interpretable document intelligence for neurological screening using hand-drawn patterns. Experimental results on publicly available Parkinson’s disease handwriting datasets demonstrate competitive predictive performance and clinically consistent explanations.

---

289. A latent dimension of Condorcet’s jury theorem for multiple AI advisers

• Authors: Kazutoshi Sasahara , Aoi Naito , Ryo Fujie
• URL: https://arxiv.org/abs/2609.14438
• Abstract:

  When the same question is asked of multiple AI advisers, as in self-consistency and LLM-as-a-judge panels, Condorcet’s jury theorem predicts that adding independent, competent advisers makes the majority more reliable. The theorem, however, has a latent dimension when viewed from the user’s vantage: adding advisers also makes disagreement more visible. A binomial model reveals that this ``visible dissent’’ becomes nearly inevitable as the number of advisers grows, and that reliability and disagreement both approach certainty but at different convergence rates. The two rates cross at an adviser accuracy of 4/5 (0.8). Below this value, visible dissent approaches certainty faster than reliability and, with enough advisers, becomes more likely than a correct majority. Even ideal panels of independent and competent advisers can be correct in aggregate but appear divided; such disagreement does not by itself indicate aggregation failure. The way advisers split also provides a common basis for predictive multiplicity, reconciliation load, and reliance miscalibration. These results indicate two distinct decisions when using multiple AI advisers: how many advisers to consult and how their verdicts should be presented and interpreted.

---

290. Lightweight Generalized DeepFake Face Detection with WAVIE: Wavelet Augmented Vision Intermediate Embeddings

• Authors: Arya Pulkit , Aditya Ruhela , Akarshan Kapoor , Arnav Bhavsar
• URL: https://arxiv.org/abs/2609.14437
• Abstract:

  Deepfake detection systems often exhibit significant performance degradation when deployed on unseen manipulation methods, limiting their reliability in real-world multimedia environments. This lack of generalization poses critical challenges for misinformation mitigation, digital forensics, and human-centric AI systems. Existing detectors perform well on the forgery methods they are trained on, but their accuracy drops sharply on unseen pipelines. To bridge this generalization gap, we propose WAVIE (Wavelet Augmented Vision Intermediate Embeddings), an end-to-end architecture that combines complementary spatial and frequency cues on top of a frozen CLIP backbone. WAVIE projects intermediate transformer embeddings through a lightweight learnable module, applies a three-level Daubechies-6 (db6) discrete wavelet transform (DWT), refines the low-frequency branch while preserving the high-frequency branch, reconstructs the feature via inverse DWT, and performs classification. Trained only on FaceForensics++, WAVIE achieves AUROC = 0.852 on Celeb-DF-v1, 0.852 on Celeb-DF-v2 and 0.831 on WildDeepFake (WDF) at the frame level, outperforming several state-of-the-art generalization baselines. Extensive ablation studies confirm the importance of both the wavelet module and the intermediate-feature aggregation for cross-dataset performance, highlighting the necessity of jointly leveraging spatial and frequency domains. These results position WAVIE as a strong baseline for deepfake detection in the wild.

---

291. Has Scientific Talent Shifted from Depth to Breadth?Evidence across Papers, Knowledge Inputs, Careers, and Teams

• Authors: Xiaoshn Nee , Haobo Zhong , Xiaomin Ni
• URL: https://arxiv.org/abs/2609.14425
• Abstract:

  Generative artificial intelligence raises a central question for scientific training and organization. Is research shifting from deep specialization toward broad individual knowledge? We examine this proposition across papers, cited knowledge, contributor histories, and teams using 47,959 articles from six fields over 2010-2025, 51,736 resolved cited works, and chronologically reconstructed prior publication histories for 1,754 randomly selected index contributors. From 2010 to 2022, team size increased by an estimated 37.3% (95% confidence interval [34.4%, 40.3%]), while paper topic breadth declined by 0.0144 on a 0-1 hierarchical distance scale. Cited knowledge was stable to modestly broader, revealing a divergence between focused outputs and the reach of knowledge inputs. Established contributors’ prior breadth increased by 0.0190 [-0.0078, 0.0459] by 2019-2022, within a +/-0.05 equivalence bound assessed in sensitivity analysis. In mature citation windows, one standard deviation of focal depth was associated with 8.2% higher 1 + FWCI [1.9%, 14.9%]; average breadth and interaction associations were smaller under the specified equivalence bounds. Post-2022 deviations from earlier trends were not systematic, and recent changes did not vary clearly with baseline AI intensity across 83 subfields. The findings support a differentiated structure of scientific expertise in which focused individual accumulation coexists with expanding collaboration and sustained access to diverse knowledge inputs.

---

292. A Generative AI Integrated Multimodal Framework for Low-Latency Multi-Camera Person Re-Identification

• Authors: Leon Fernando , C Dombawala , P. Hettigoda , Vanodhya G. Warnasooriya , Ishara Neranjana , Rashmika Nawaratne
• URL: https://arxiv.org/abs/2609.14419
• Abstract:

  Person re-identification (ReID) is essential for multi-camera surveillance and tracking, yet remains difficult due to viewpoint and illumination changes, occlusion, background clutter, and low resolution imagery. We propose a generative AI integrated multimodal ReID framework designed explicitly for robustness under missing cues and low latency deployment. The key idea is a cost aware early-exit cascade that prioritizes inexpensive, high confidence evidence and only triggers expensive modalities for ambiguous cases. Our system integrates (i) global visual embeddings from segmented person regions, (ii) automatically generated fine grained semantic attribute descriptions generated by vision-language models (VLMs), and (iii) optional facial embeddings when face observations are reliable. To optimize the balance between accuracy and latency, we use a cost aware early-exit cascade instead of fusing all modalities. Specifically, we first inspect the top-k retrieval results to determine whether the query is unambiguous. If the best match is clearly separated from the remaining candidates, we stop early and return the result to minimize latency; in ambiguous cases, we keep multiple hypotheses and invoke additional modalities (face/semantic) with adaptive reliability weighting to refine the decision. We report person re-identification performance using mAP and Rank-1 accuracy on the Market-1501 and DukeMTMC-reID benchmarks. The proposed adaptive early-exit cascade resolves 60.7% of DukeMTMC-reID queries and 68.4% of Market-1501 queries without invoking semantic reasoning, reducing computational overhead while maintaining competitive retrieval performance.

---

293. Surrogate-Assisted Genetic Programming with Phenotypic Characterisation in Dynamic Multi-Mode Project Scheduling

• Authors: Yuan Tian , Yi Mei , Mengjie Zhang
• URL: https://arxiv.org/abs/2609.14418
• Abstract:

  Dynamic multi-mode resource-constrained project scheduling requires decisions to be made under precedence constraints, limited resources, multiple execution modes, and uncertain activity durations. Genetic programming (GP) can automatically evolve heuristic rules for such problems, but its simulation-based fitness evaluation is computationally expensive. This study investigates phenotypic characterisation (PC) in surrogate-assisted GP to evolve higher-quality scheduling heuristics under a fixed budget of full simulation-based fitness evaluations. A key question is how GP individuals should be encoded into phenotypic characterisations to support effective fitness estimation. To answer this question, three PC encoding schemes with different levels of information richness are designed: priority-value encoding, which preserves raw rule outputs; rank encoding, which captures candidate ordering; and binary encoding, which represents final scheduling decisions. These encodings are combined with different distance metrics to measure behavioural similarity between GP individuals. The experimental results show that binary encoding with Euclidean distance provides the most effective and robust surrogate guidance. Further analyses show that surrogate estimation accuracy alone does not fully explain the performance differences. The PC representation also determines how effectively phenotypically redundant offspring are removed and how much behavioural diversity is retained after preselection. Ablation experiments further demonstrate that duplicate removal and surrogate preselection provide complementary benefits, with their combination producing the largest improvement. These findings highlight that effective surrogate-assisted GP depends not only on identifying promising offspring, but also on controlling redundancy and preserving useful diversity during evolutionary search.

---

294. A Hybrid Dependency-Aware Framework for Task Decomposition and Dynamic Agent Generation in Oracle-to-PostgreSQL Migration

• Authors: Oleg Grynets , Oleg Kaskun , Alona Seletska , Daryna Tukalo , Vasyl Lyashkevych
• URL: https://arxiv.org/abs/2609.14413
• Abstract:

  Large language model (LLM)-based database migration is often treated as direct code transformation, although enterprise Oracle systems contain heterogeneous SQL and PL/SQL artifacts with different dependencies, execution order, complexity, and validation needs. This paper proposes a hybrid dependency-aware framework that identifies migration tasks, builds a cross-file dependency graph, condenses cyclic dependencies, and uses task specifications to generate specialized migration agents at runtime. The deterministic path combines ANTLR-based parsing with typed dependency extraction, while an LLM fallback is invoked only for units that cannot be parsed reliably. On a corpus of 116 Oracle files, the pipeline produced 1,037 units with zero coverage gaps and 1,271 AST-derived dependencies. The fallback processed 165 parse-error units, recovered 496 additional validated dependencies, eliminated unresolved-dependency units, and increased resolved internal edges from 446 to 527. The graph contained four dependency-respecting phases, while cycle handling was validated separately using Tarjan SCC condensation. A complementary experiment on 1,006 PL/SQL files regenerated 623 scripts (~62%), of which 380 (~61%) executed successfully in PostgreSQL 16. Tables achieved about 85% regeneration success, whereas no query regenerations succeeded under the evaluated specification-mediated baseline, and procedural objects remained strongly dependent on schema context. These results motivate dependency-aware context delivery, task-specific validation, and differentiated agent routing. The paper also formalizes task-to-agent mapping, introduces monitoring and diagnostics for execution feedback, and defines a controlled comparison of monolithic, static-decomposition, dependency-aware, and dynamically orchestrated migration strategies.

---

295. LLaTSA: Large Language Model-Aligned General-Purpose Transient Stability Analysis

• Authors: Chao Shen , Hongwei Zhen , Junyan Shao , Zhenghao Yang , Yifan Zhang , Mingyang Sun
• URL: https://arxiv.org/abs/2609.14374
• Abstract:

  Dynamic trajectory prediction has become an important paradigm for data-driven transient stability analysis (TSA), yet most existing predictors remain system-specific and require substantial retraining when network configurations, generation mixes, or state-variable sets change. Uni-TSA introduced a general-purpose TSA framework that combines channel-independent modeling with a pretrained large language model (LLM) predictor. Nevertheless, its application to heterogeneous systems is limited by ambiguity in short observations, a mismatch between numerical trajectories and LLM embeddings, neglected coupling among state variables, and the high inference cost of dense backbones. This paper proposes LLaTSA, an LLM-aligned framework for general-purpose trajectory-based TSA. LLaTSA first incorporates operating conditions, disturbance attributes, and state-variable identity through a structured textual prefix. It then aligns normalized temporal patches with a TSA-related vocabulary before processing them with a pretrained sparse decoder-only mixture-of-experts (MoE) backbone. A state-variable coupling module captures coordinated post-fault evolution, while teacher forcing and rollout-based training support iterative long-horizon prediction. Case studies on multiple test systems demonstrate accurate trajectory prediction, reliable stability discrimination, and effective adaptation across unseen scenarios.

---

296. AURA: Unified Multimodal Framework for Conversational Music Editing

• Authors: Quoc-Huy Trinh , Minh-Van Nguyen , Debesh Jha
• URL: https://arxiv.org/abs/2609.14344
• Abstract:

  Instruction-guided music editors typically process each request independently, limiting their ability to support workflows in which users progressively refine a track. We introduce AURA, a unified multimodal framework for conversational music editing. AURA uses a multimodal large language model to interpret the complete dialogue history, an optional image, and reference audio, distilling the editing intent into compact concept tokens. A concept-to-audio module injects these tokens and frame-aligned reference features into a frozen MusicGen backbone, enabling precise edits while preserving unaffected content. AURA optimizes only 91M parameters while retaining 1.9B frozen backbone parameters. Experiments on Slakh2100 and MoisesDB demonstrate substantial improvements in edit correctness and content preservation over existing instruction-guided methods, including a 4-5 times reduction in FAD for out-of-domain addition and removal.

---

297. Biquaternionic Space with Complex-valued Attention for Temporal Knowledge Graph Completion

• Authors: Rushan Geng , Cuicui Luo
• URL: https://arxiv.org/abs/2609.14279
• Abstract:

  Temporal knowledge graph embedding (TKGE) models infer missing facts in knowledge graphs that evolve over time. Many existing models use a single geometric space, which can limit their ability to represent diverse relational patterns, or treat entity representations as static. We propose Biquaternionic Space with Complex-valued Attention (BSCA), a TKGE model that combines circular and hyperbolic rotations within a unified biquaternionic framework. A complex-valued attention mechanism adaptively fuses time-conditioned and relation-conditioned entity representations, allowing them to vary with temporal and relational context. Experiments on five benchmark datasets show competitive performance across datasets, with the largest improvement on GDELT: BSCA achieves an MRR of 52.1\%, compared with 38.1\% for the strongest baseline in our comparison.

---

298. SpermYOLO: A Coordinated YOLO-Based Detector for Accurate and Efficient Sperm and Impurity Detection in Microscopic Images

• Authors: Shengqi Chen , Zilin Wang , Xingyu Pan , Wenting Yu , Pengchao Deng , Guohua Wu
• URL: https://arxiv.org/abs/2609.14278
• Abstract:

  Accurate sperm detection is essential for computer-assisted semen analysis, yet it remains challenging in microscopic images due to dense distributions, visually similar artifacts, and sperm-like impurities. In this paper, we propose SpermYOLO, a coordinated and compact YOLOv11-derived framework for joint sperm and impurity detection in microscopic images. SpermYOLO introduces four architectural improvements: C3k2-IDB for channel-wise discriminative feature extraction, D2SEM for spatial–spectral semantic enhancement, MFM for adaptive multi-scale feature fusion, and the DESD Head for detail-enhanced shared prediction. Experiments on the SVIA semen microscopic imaging benchmark show that SpermYOLO achieves 97.2\% sperm AP and 75.4\% impurity AP, outperforming generic detectors, dedicated sperm detection models, and improved YOLO variants. Compared with the baseline model, SpermYOLO improves sperm AP, impurity AP, $\mathrm{mAP}{50}$, and $\mathrm{mAP}{50:95}$ by 1.6, 10.0, 5.8, and 2.7 percentage points, respectively, while preserving a lightweight model scale. Cross-scene evaluation on the SDTB testicular-biopsy microscopy benchmark shows that SpermYOLO remains effective with extremely small sperm targets and complex tissue backgrounds, achieving the highest $\mathrm{mAP}{50}$ and $\mathrm{mAP}{50:95}$ of 74.8\% and 31.2\%, respectively. Ablation studies and qualitative analyses further support these improvements by demonstrating the contributions of the proposed modules and showing more focused feature response patterns than the baseline model. These findings suggest that SpermYOLO is an effective and efficient approach for sperm detection in challenging microscopic imaging scenarios.

---

299. ATTRICITE: Training an Open 4B Model for Citation Recovery toward Faithful Attribution

• Authors: Yee Man Choi , Xuehang Guo , Songcheng Cai , Yimu Wang , Yi R. Fung , Qingyun Wang
• URL: https://arxiv.org/abs/2609.14248
• Abstract:

  Faithful citation attribution begins with identifying the intended source for a scientific claim. We study this source-identification capability through citation recovery: recovering the paper cited by the original author from a citation-bearing passage. Our evaluation adopts the published author’s citation as an observable human attribution signal and uses target recovery as a proxy for progress toward faithful attribution. We introduce ATTRICITE, an open 4B-parameter model trained for tool-using citation recovery within the CiteGuard retrieval environment, together with CITEALIGN, a 7,607-instance computer-science dataset drawn from recent scientific literature. For controlled evaluation, we construct a 709-instance benchmark subset of CITEALIGN, comprising 410 development instances from 2024 publications and 299 temporally held-out test instances from 2025 publications. Across three runs at an inference temperature of 0.7, GRPO fine-tuning improves Qwen3-4B from 49.4%$\pm$1.5% to 59.8%$\pm$0.2% target-match accuracy, a gain of 10.4 percentage points. Despite using only 4B parameters, ATTRICITE outperforms gpt-oss-20b and comes within 3.9 points of GPT-5.4-mini, while Gemma 4 31B IT achieves the strongest overall performance at 72.0%$\pm$1.0%. We release the model and collection pipeline this https URL to support reproducible research on citation recovery toward faithful attribution in a continually evolving scientific literature.

---

300. The Attribution-Compression Frontier in Retrieval-Augmented Generation

• Authors: Deepanshu Mody
• URL: https://arxiv.org/abs/2609.14245
• Abstract:

  Context compression reduces generator input in retrieval-augmented generation, but answer quality alone does not characterize citation attribution. We measure citation attribution across compression methods and budgets, comparing reranking, extractive selection, abstractive summarization, token pruning, and an extract-cluster-rewrite construction on ASQA and QASPER under a fixed generator and primary entailment evaluator. On ASQA at a nominal 0.25 budget (achieved compression 0.08), a RECOMP-style compressor’s citations score 0.86 precision against its summaries but 0.12 against source spans under our re-attributability protocol. These estimates depend on a shared NLI model for span recovery and citation scoring and lack independent human calibration. Extractive selection’s observed grounded precision ranges from 0.43 to 0.49 across nominal budgets from one-half to one-tenth of the ASQA context, while answer quality declines. For the same RECOMP setting, claim verification after source recovery yields an unsupported rate of 0.88 versus 0.17 when checking summaries. This gap persists beyond structural rejection of missing provenance, but remains evaluator-dependent. A 200-question TRUE T5-XXL audit also finds emitted–grounded gaps under both fixed and recomputed source mappings, without establishing human-calibrated support rates.

---

301. OpWeave: Flexible Operator Disaggregation for Heterogeneous LLM Serving

• Authors: Zikun Li , Yixuan Mei , Shiqi Pan , Zixuan Chen , Xiaowen Zhang , Mengdi Wu , Shuhuai Lin , Yutong Yang , Zhihao Zhang , Xupeng Miao , Rashmi Vinayak , Zhihao Jia
• URL: https://arxiv.org/abs/2609.14237
• Abstract:

  LLM serving systems increasingly disaggregate inference into finer-grained stages, with recent approaches separating attention from FFN or MoE execution during decode. This operator-level disaggregated serving (ODS) can improve hardware matching and enable independent scaling, particularly across heterogeneous devices. However, existing systems fix operator boundaries and lack a unified characterization of when disaggregation reduces serving cost. We present OpWeave, an end-to-end framework for heterogeneous ODS. OpWeave provides an analytical cost model that bounds the gains of homogeneous and heterogeneous ODS over colocated serving. It jointly optimizes operator partitioning and deployment configuration through a regularity-aware planner that keeps the search tractable even for hybrid-attention models. A vLLM-based runtime executes the synthesized plans with flexible operator stages across heterogeneous device groups. In our evaluation, OpWeave reduces serving cost by up to $1.78\times$ on homogeneous and $1.89\times$ on heterogeneous GPU clusters relative to the best feasible baseline, while meeting latency SLOs.

---

302. Assessing the Applicability of Existing Design Recommendations to AI Companion Design: A Multi-Method Study

• Authors: Soobin Cho , Deveshi Modi , Divya Mavinkurve , Jieqiong Ding , Mark Zachry
• URL: https://arxiv.org/abs/2609.14236
• Abstract:

  With the rapid proliferation of large language model (LLM)-based systems, AI companions have emerged as conversational agents designed to cultivate emotional connection rather than primarily to support humans in instrumental tasks. Because engagement with AI companions involves relational, emotional, and potentially long-term interactions, their design is consequential. Prior work has offered guidance for designing trustworthy and relational AI systems and has begun to examine design for AI companionship. However, while such work provides insights into possible design solutions, less is known about what makes AI companion design difficult as a design problem. To examine this challenge, we assessed the applicability of existing design recommendations from adjacent domains in the context of AI companion design. Our multi-method investigation unfolded across four phases: literature review, practitioner co-analysis, internal heuristic evaluation, and external expert assessment. Throughout this process, we synthesized nine design principle areas that surfaced tensions in the applicability of existing recommendations to AI companion design. Our findings show that ethical and UX-oriented considerations are deeply intertwined and often require context-sensitive application. We document a systematic, multi-method problem analysis that uses these principle areas as an analytic artifact to examine why existing recommendations cannot be directly transferred to AI companion contexts.

---

303. Graph-Transformer Fraud Detection with Self-Supervised Pretraining and Conformal Risk Control

• Authors: Sergei , Komarov
• URL: https://arxiv.org/abs/2609.14234
• Abstract:

  Financial fraud in corporate transaction networks has grown more coordinated and harder to detect with rule-based engines and with classical learning models that treat each transaction in isolation. This paper presents GTFD, a graph-transformer fraud detector that fuses structural and temporal evidence from a corporation’s payment graph. GTFD encodes the graph with a multi-head graph attention network, encodes ordered transaction sequences with a gated transformer, and combines both views through a cross-modal gating layer. A conformal risk-control head converts the fused representation into threshold-free anomaly scores with finite-sample coverage guarantees, and the network is trained with self-supervised link-mask pretraining plus adversarial augmentation so it remains stable under scarce labels and under adversarial perturbation. On a corporate transaction benchmark enriched with coordinated fraud rings, GTFD reaches an AUROC of 0.990, an F1-score of 96.1% (precision 96.3%, recall 95.9%), and an accuracy of 98.4%. It reduces the false-positive rate by about 29% relative to the strongest baseline while raising coordinated fraud-ring recall from 85.1% to 96.5%. Ablations attribute roughly 2.0 AUROC points to self-supervised pretraining and 1.9 AUROC points to the conformal head, and adversarial stress tests show GTFD retains 89.2% accuracy at perturbation magnitude 0.20 where the next-best model falls to 76.4%.

---

304. Modeling, Scaling, and Decoding: Optimizing Controllable Speech Generation with Nonverbal Vocalizations

• Authors: Ziyu Zhang , Yun Chen , Taihui Wang , Hanzhao Li , Qicong Xie , Rilin Chen , Zhixian Zhao , Lei Xie
• URL: https://arxiv.org/abs/2609.14231
• Abstract:

  Controllable synthesis of nonverbal vocalizations (NVVs) is es- sential for natural and expressive speech, but remains challeng- ing due to their acoustic diversity and imbalanced distribution in existing corpora. To address these challenges, we develop an NVV-aware DiTAR system that models continuous speech latents, encodes the 16 target NVV categories as dedicated to- kens, and adapts stop prediction to distinguish mid-utterance vocalizations from utterance boundaries. Training begins with large-scale bilingual pre-training on diverse NVV speech, fol- lowed by continued supervised fine-tuning on a corpus en- hanced through targeted synthetic augmentation and frequency- aware rebalancing. At inference time, we select the acoustic prompt, tune the LM-guidance and noise-injection scales, and apply Best-of-N sampling with multi-metric selection to re- duce generation failures. The final system achieves an official weighted bilingual score of 62.786, ranking first in Mandarin, second in English, and first overall among participating systems in Track 2 of the ISCSLP 2026 NVVSpeech Challenge. Ab- lation studies show that targeted augmentation benefits under- represented NVV categories the most, while robust candidate selection requires balancing NVV correctness, lexical fidelity, and perceptual quality.

---

305. Task-Specified Active Metrological Inspection with Measurement-Steered VLA Manipulation and Deterministic Evidence Gating

• Authors: Zhiling Chen , Jingzhan Ge , Ruimin Chen , Matthew P. Castanier , David Gorsich , Farhad Imani
• URL: https://arxiv.org/abs/2609.14219
• Abstract:

  High-mix low-volume (HMLV) manufacturing requires inspection systems to adapt to changing parts, specifications, and work orders without repeated task-specific programming. Existing inspection automation typically assumes predefined sensing sequences, while general purpose robot agents optimize task completion rather than the completeness and validity of metrological evidence. We formulate task-specified active metrological inspection and propose From Requirements to Admissible Metrological Evidence (FRAME), a hierarchical dual-arm framework that converts an inspection instruction and structured specification into traceable conformance evidence. FRAME coordinates learned manipulation with calibrated laser profilometry: a task manager grounds and schedules requirements, active surface correspondence verifies physical-to-specification localization, and evidence memory tracks measurement provenance, admissibility, and coverage. Learned components may propose inspection targets and physical access actions, but deterministic datum-grounded measurement, admissibility checks, coverage auditing, and conformance evaluation prevent incomplete or unverified evidence from authorizing PASS. A series of physical experiments shows that FRAME achieves higher end-to-end inspection reliability, fewer false accepts, and shorter task completion time.

---

306. ECAS: An Edge-Controlled Agentic System for Validation-Gated Scientific Application Execution

• Authors: Baixi Sun , Mingze Xia , Huihuo Zheng
• URL: https://arxiv.org/abs/2609.14211
• Abstract:

  Scientific applications increasingly rely on high-performance computing (HPC), yet translating a scientist’s high-level goal into a correct target-scale execution remains brittle and labor-intensive. Large language model (LLM) agents promise to automate this, but two obstacles remain: granting a cloud-hosted model direct HPC access exposes credentials and execution authority, while withholding it demands continuous human supervision; and one-shot generation cannot adapt when generated artifacts fail in a site-specific HPC environment. We present \textsc{ECAS}, an \textbf{E}dge-\textbf{C}ontrolled \textbf{A}gentic \textbf{S}ystem for closed-loop execution of scientific computing campaigns with limited human intervention. \textsc{ECAS} separates \emph{reasoning}, \emph{control}, and \emph{execution}: a cloud-hosted LLM proposes plans, artifacts, and repairs; a user-controlled edge agent retains credentials, workflow state, and execution authority while enforcing policy and resource constraints; and the HPC system computes. Its core mechanism is \emph{validation-gated execution}: generated artifacts pass static checks and small-scale validation, failures trigger repairs from sanitized execution feedback, and target-scale execution is permitted only after validation and policy checks pass. \textsc{ECAS} also draws on an edge-resident library of expert-distilled, site-specific skills that is never disclosed to the cloud. In preliminary experiments with three scientific applications on two production ALCF systems under six injected fault types, closed-loop repair improves application success from 0/6 to 6/6 over one-shot generation, validation gating prevents all three observed target-scale failures, and skill conditioning improves success from 4/6 to 6/6. These results show the feasibility of delegating adaptive reasoning to the cloud while retaining execution control at the edge.

---

307. Data-free On-policy Distillation

• Authors: Gengsheng Li , Mao Zheng , Mingyang Song , Jie Sun , Zeyuan Liu , Ruiqi Liu , Qiyong Zhong , Haiyun Guo , Junfeng Fang , Jinqiao Wang
• URL: https://arxiv.org/abs/2609.14193
• Abstract:

  On-policy distillation (OPD) has become a standard component of frontier post-training pipelines, yet how much its training data actually contributes has gone largely unexamined. On the two teacher-student pairings most common in practice, we find OPD almost indifferent to its data: eight prompts already match a 17k-problem dataset, and three independently built datasets whose difficulty and teacher-student KL differ several-fold produce nearly indistinguishable training curves. Two causes account for this. First, the unit of data in OPD is the state a prompt leads to, not the prompt itself: a single prompt keeps exposing new teacher correction as sampling continues, while the marginal value of additional prompts collapses after eight. Second, replacing mathematics with competitive programming still recovers over ninety percent of the in-domain gain, indicating that OPD transfers the teacher’s mode of reasoning rather than knowledge related to the data. We take this to its limit with Data-free On-policy Distillation (DF-OPD), in which the teacher writes its own training questions under a simple prompt – no external data, no filtering – leaving a system of just two policies. DF-OPD matches and even surpasses real data, and the questions it produces track the teacher’s own post-training data on three key diagnostics of training dynamics, which other real datasets do not. Applied to multi-teacher distillation, where the (prompt, domain) pairs normally have to be derived from post-training data that is often out of reach, 1k self-generated questions close 98.5% of the available headroom, even surpassing the 96.6% reached with 7k real examples. Moreover, together these results invite a reassessment of the role data plays in OPD.

---

308. Entropy-Punctured Bloom Filters for Memory-Efficient Machine Learning

• Authors: John Cartmell , Mihaela Cardei , Ionut Cardei
• URL: https://arxiv.org/abs/2609.14187
• Abstract:

  Memory-efficient feature representations are increasingly important in machine learning settings where storage, transmission cost, bandwidth, or privacy constraints limit access to raw data. Bloom Filter (BF) encodings provide compact probabilistic representations of engineered features, but their behavior under structural compression and their applicability to regression tasks remain underexplored. In this work, we propose entropy-punctured Bloom Filters, a memory-aware encoding strategy that removes low-variability bit positions identified using empirical entropy. Starting from fixed-length BF encodings of quantized features, the proposed approach produces reduced representations that preserve predictive structure while improving predictive efficiency relative to encoded representation size. We evaluate the approach on diverse regression datasets, comparing raw features, Principal Component Analysis (PCA), Random Projection (RP), and Bloom Filter variants under leakage-free evaluation protocols and approximately matched representation sizes. Performance is assessed using ridge regression, XGBoost, and neural networks, with predictive efficiency measured as R2 relative to encoded representation size per sample. Results show that Bloom Filter encodings remain competitive with classical compressed representations while achieving substantial storage savings. Entropy-based puncturing further reduces representation size with minimal loss in predictive fidelity, yielding improved predictive efficiency. These findings demonstrate that entropy-punctured Bloom Filters provide an effective representation-level compression approach for memory-constrained machine learning.

---

309. Bi-Level Routing and Sparse Spatial Attention based Multi-View BEV 3D Object Detection for Autonomous Driving

• Authors: Jing Zhang , Jiaqi Liu , Zibo Wang
• URL: https://arxiv.org/abs/2609.14185
• Abstract:

  Bird’s Eye View (BEV)-based multi-view 3D object detection suffers from challenges of computational complexity, multi-scale feature extraction, and efficiency of dense 2D-to-BEV view transformation. To address these problems, this paper proposes an improved BEV 3D object detection algorithm Sparse-BEVNet. Firstly, a Bi-Level Routing Attention (BRA) mechanism is introduced into the image feature extraction network to reduce the computational burden of the backbone. Second, Cascaded Group Attention (CGA) is employed in the feature fusion module, which enhances deep interaction across features of different hierarchical levels without introducing additional computational overhead. Furthermore, a Sparse Spatial Cross-Attention mechanism is adopted to replace the conventional dense view projection pipeline. Experimental results on the public nuScenes dataset demonstrate that the proposed method achieves a mean Average Precision (mAP) of 45.2% and a nuScenes Detection Score (NDS) of 54.5%, corresponding to 3.6% and 2.8% improvements relative to the baseline model, respectively.

---

310. Towards Evolving Context Parameterization for Large Language Models

• Authors: Xiaobing Shi , Zherui Li , Yiming Jiang , Kun Wang , Yufei Guo
• URL: https://arxiv.org/abs/2609.14168
• Abstract:

  Context parameterization enables large language models (LLMs) to internalize contexts into reusable model parameters, avoiding repeated processing across subsequent queries. However, existing methods typically assume static contexts and lack explicit mechanisms for distinguishing validity states under continual updates. To study this real-world scenario, we formalized the Memory Updating with Sequential Evolution (MUSE) task and constructed MUSE-bench to evaluate update incorporation and unaffected-information preservation. The resulting challenge requires preserving the global state while adjusting the contribution of memory evidence. Motivated by this, we proposed PLUME, a training-free method that constructs a global update representation, activates memory evidence to form a local parameter view, and adaptively integrates their predictions during decoding. Comprehensive evaluation on MUSE-bench demonstrated PLUME’s effectiveness in sequential evolution settings, yielding relative improvements of 29.9% in average ROUGE-L Recall and 54.9% in LLM-as-a-Judge. Our codes are available at: this https URL .

---

311. Signatures of Steerability in Activation Space of Language Models

• Authors: Prajjwal Bhattarai , Tuka Alhanai
• URL: https://arxiv.org/abs/2609.14151
• Abstract:

  Steering language models using a set of contrastive representations has been a canonical and computationally efficient method for controlling model behavior. Despite this success in controlling certain model behaviors, the effectiveness of activation steering varies markedly across concepts; the generalization properties of steering vectors are often considered a function of the dataset used to construct them. We make this dataset-dependence claim more rigorous and show that simple separation metrics strongly correlate with the downstream steerability of language models across diverse settings, even after controlling for layers and dataset effects. Beyond prediction, we provide evidence from a synthetic superposition experiment that separation metrics are strongly correlated with alignment between the empirical and true feature direction. Our results suggest that simple separability statistics can serve as practical diagnostics for when steering vectors are likely to work.

---

312. A New Transformer-Based Approach for Audio-Based Kinship Verification and a New Uncontrolled Mandarin Kinship Speech Dataset

• Authors: Qiyang Sun , Langqing Zhang , Yupei Li , Björn Schuller
• URL: https://arxiv.org/abs/2609.14145
• Abstract:

  Kinship verification is a task involving determining whether two individuals share a first-order kin relation. To tackle this task, we propose CONVTRAP-TN, a new architecture for audio-based kinship verification, and conduct an ablation study on the proposed model. To the best of our knowledge, we are the first to apply the successful transformer architecture to the task of audio-based kinship verification. Furthermore, we also collect a custom speech dataset, ARKIN, which accurately reflects everyday recording conditions. We do this because only a few speech datasets with kinship labels currently exist, all of which either source extremely noisy in-the-wild data from the internet, or instruct speakers to record in specific environments. These settings fail to reflect real-world scenarios where users record on personal devices under unrestrained conditions. Additionally, we perform a series of preliminary baseline experiments on the collected dataset, including speaker verification and recognition, speech recognition, age estimation, and kinship verification, as well as cross-dataset kinship verification experiments to show that existing methods are not robust across datasets.

---

313. LIMBO: Lifelong Inference-Time Memory and Budget Optimization for LLM Agents

• Authors: Siddharth Sharma , Nilesh Prasad Pandey , Onat Gungor , Tajana Rosing
• URL: https://arxiv.org/abs/2609.14138
• Abstract:

  As LLM agents become integrated into increasingly complex workflows, they must continually acquire new capabilities while retaining competence on previously learned tasks. Lifelong agents address this through experience replay, injecting past interactions into the prompt to leverage prior experience during inference. However, replay is not free: every replayed trajectory competes with retrieval, reasoning, tool use, and verification for the same limited prompt and compute budget, making effective resource allocation essential. Existing approaches allocate these resources using fixed replay policies, regardless of whether replay is beneficial for the current task. We identify this as inference-time memory allocation, a distinct problem class for lifelong agents, and introduce LIMBO: the first online framework to our knowledge that treats memory as a controllable inference-time resource and jointly optimizes memory strategy and inference budget for each incoming task. Unlike prior approaches that fix the replay policy or require model weights, teacher supervision, or offline retraining, LIMBO learns this allocation online in a single pass, explicitly balancing task performance and inference cost without modifying the underlying agent. Across three LLM backbones on LifelongAgentBench, LIMBO achieves better cost-accuracy tradeoffs than state-of-the-art memory-augmented baselines and nearly matches all strongest such baselines at up to ~83% lower inference cost (~53% on average). LIMBO adapts its policy across models and environments without retraining, demonstrating that effective allocation can be learned online rather than manually specified.

---

314. To do($x$) or not to do($x$): Medical Image Counterfactuals for Dataset Augmentation

• Authors: Yasin Ibrahim , Robin J. Evans , Konstantinos Kamnitsas
• URL: https://arxiv.org/abs/2609.14124
• Abstract:

  Medical image analysis is often hindered by biased datasets, which can lead to biased models and limited clinical applicability. A promising strategy for mitigating such biases is to augment training data with synthetic images. Counterfactual (CF) generation is one such strategy, though the term is used in two different senses: in some works, CFs are produced through causality-based interventions derived from structural causal models, whereas in others, they are produced by non-causal image edits or conventional conditional generative models, such as altering anatomy or adding pathologies. In this work, we study this distinction and evaluate its practical consequences for medical image augmentation. We compare three conditioning strategies: $\textit{Deterministic}$, which changes selected variables while holding the remaining variables fixed; $\textit{Undirected}$, which updates variables according to learned statistical associations without assigning causal directions; and $\textit{Causal}$, which propagates interventions along a directed causal graph. We analyse how these choices affect the resulting images, and explore when causally grounded methods improve dataset augmentation or bring limited benefit. In particular, we assess downstream performance and fairness, where fairness refers to reduced sensitivity to dataset biases across sensitive subgroups. Our experiments demonstrate that using a causal approach to synthetic training data generation can lead to tangible benefits, with these insights offering valuable guidance to machine learning practitioners for the effective design of data generation protocols.

---

315. Talking to Me or Someone Else? Rethinking Talk-to-Me Detection in Egocentric Videos

• Authors: Feiyu Du , Xi He , Jia Li , Yapeng Tian , Weili Wu
• URL: https://arxiv.org/abs/2609.14118
• Abstract:

  Online understanding of who is talking to the camera wearer is a key capability for egocentric social interaction. However, existing talk-to-me (TTM) studies are commonly formulated as offline clip-level recognition, which is poorly aligned with online interaction and overlooks the diverse non-TTM speaking states that naturally arise in egocentric videos. In this paper, we revisit this problem by reformulating it as an online, frame-level prediction task. Instead of treating TTM as a binary problem against a single negative class, we model it in the presence of diverse and previously underexplored non-TTM states, such as talking-to-others, self-talking, and background conditions. To support this new formulation, we construct an Online TTM Dataset consisting of 406 egocentric video clips with approximately 900K annotated frames, each labeled with frame-level social interaction categories (e.g., background, TTM, talking-to-others, self-talking), by extending the Ego4D social interaction benchmark. In this benchmark, we evaluate five adapted baselines and develop a new model that integrates social cues across modalities. Experimental results show that our multimodal model, which jointly leverages audio, visual, and speech-semantic cues, achieves 75.5% frame-level F1 on TTM, outperforming strong baselines and enabling a systematic analysis of how different speaking states affect TTM recognition.

---

316. Real-Time Synthesis of Robust Controlled Invariant Sets for Monotone Systems

• Authors: Yasin Sonmez , Mahmoud Khaled , Majid Zamani , Murat Arcak
• URL: https://arxiv.org/abs/2609.14115
• Abstract:

  Safety-critical control of autonomous systems requires formal safety certificates, such as controlled invariant sets, that must be computed online as conditions change. Although standard synthesis algorithms scale poorly with state dimension, monotone dynamical systems with lower-closed safety specifications allow for accelerated computation of controlled invariant sets. In particular, lazy fixed-point algorithms exploit monotonicity and track only the antichain basis of the set. However, membership tests and redundancy checks against an evolving basis remain major bottlenecks. We introduce a threshold-function reformulation in which a lower-closed set on a d-dimensional grid is represented by its column heights along a designated axis. This reformulates the greatest-fixed-point iteration as independent one-dimensional binary searches, one per grid column, yielding an embarrassingly parallel iteration with asymptotically lower computational complexity than the lazy fixed-point algorithm. Experiments synthesize invariant sets on 3D grids with 10^9 cells in under 50 ms and 10^14 cells in under two minutes. We further demonstrate online re-synthesis in a safety-informed model predictive controller example.

---

317. A Voxel-Spacing-Aware Extension of PyRadiomics for Anisotropic Texture Analysis

• Authors: David Corral Fontecha , Juan Miranda Bautista , Pablo Menendez Fernández-Miranda , Andrea Trapote Fernandez , Lara Lloret Iglesias , Jose A. Vega
• URL: https://arxiv.org/abs/2609.14103
• Abstract:

  Radiomic texture features are commonly extracted from anisotropic CT and MRI acquisitions, where identical voxel offsets may represent different physical distances. We implemented and validated a voxel-spacing-aware extension of PyRadiomics that incorporates spacing information without generating interpolated gray levels. The framework operates across the Python frontend, C wrapper, and computational backend. GLCM uses anisotropy-relative feature-level angular aggregation, NGTDM uses anisotropy-relative weighted neighborhood averaging, and GLRLM, GLDM, and GLSZM are computed on a finite-volume zero-order-hold representation derived from the native anisotropic grid. Synthetic 3D phantoms were used for software validation. The modified implementation reproduced standard PyRadiomics exactly when spacing-aware mode was disabled and remained equivalent under isotropic spacing across 75 texture features. Under anisotropic spacing, the method selectively modified texture families and was numerically distinct from nearest-neighbor, linear, and B-spline resampling. Computational profiling showed moderate runtime and memory increases, while sensitivity analyses quantified finite-volume rounding effects and confirmed that spacing-aware differences persisted across binWidth settings. The framework provides a backward-compatible technical basis for future evaluation of spacing-aware radiomics in heterogeneous medical imaging datasets.

---

318. RA-CoA: Training-free Fashion Image Captioning via Retrieval-Augmented Chain-of-Attributes

• Authors: Abhirama Subramanyam Penamakuri , Shreya Shukla , Anand Mishra
• URL: https://arxiv.org/abs/2609.14100
• Abstract:

  Fashion Image Captioning (FIC) plays a vital role in enhancing user experience and product search in e-commerce platforms. Unlike natural scene image captioning, FIC requires fine-grained visual reasoning and knowledge of domain-specific terminology to capture subtle attributes such as neckline and closure types, graphic patterns, and dress silhouettes. Moreover, as fashion inventories evolve rapidly with new trends, styles, and frequently emerging vocabulary, developing training-free captioning solution becomes essential for scalability and real-world adaptability. Instruction-tuned vision-language models (VLMs) offer a promising solution to fashion image captioning dueto their strong zero-shot capabilities and natural language fluency. However, these general-purpose models often lack attribute-level coverage and precision, and tend to hallucinate or misidentify fine-grained fashion details, making them less suitable for high-fidelity applications like product cataloging or personalized recommendations. To address this, we propose RA-CoA (Retrieval-Augmented Chain-of-Attributes), a novel, training-free framework that disentangles fashion image captioning into two interpretable stages: (i) retrieval of relevant attribute sets from a product knowledge base, and (ii) attribute-level reasoning to generate the final caption. RA-CoA is a model-agnostic approach that works with frozen VLMs to improve fine-grained attribute precision in product captions without the need for fine-tuning. Extensive evaluations across diverse VLM model families under different prompting paradigms demonstrate that RA-CoA significantly improves caption quality, achieving an average gain of 26.3% METEOR score over zero-shot captioning. We make our code publicly available.

---

319. LPA-CWM: A Learned Physical Adjudicator for Motion Reasoning with Counterfactual World Models

• Authors: Kunwei Wu , Xiang Liu , Guocai Yao , Junming Chen , Zhikang Chen , Min Zhang , Pengwei Wang , Sen Cui
• URL: https://arxiv.org/abs/2609.14073
• Abstract:

  Counterfactual world models (CWM) extract motion from pretrained video predictors by comparing factual and intervened predictions. However, responses generated under different target-frame masks vary in reliability, while uniform aggregation weights them equally. We formulate response aggregation as candidate reliability learning and propose LPA-CWM with a lightweight Learned Physical Adjudicator (LPA). Trained on dense MOVi-F trajectories, the 3.0M-parameter LPA compares visual context and response structure across an unordered candidate set to predict relative weights, while the CWM predictor and intervention generator remain frozen. The weighted responses undergo windowed localization and one paired re-evaluation to recover motion. We also introduce Completeness-aware Motion Correspondence (CMC), a ground-truth-anchored evaluation protocol that jointly measures localization, trajectory completeness, visibility, and continuity, counting missing predictions as failures on visible dynamic points. On the evaluated DAVIS and Kinetics subsets, LPA-CWM improves $\mathrm{DCA}_{\mathrm{avg} }$ over Uniform CWM by 60.0\% and 29.0\%, respectively, and also improves tracking accuracy under TAP-Vid First. A quick overview is available at this https URL .

---

320. GraMRAG: Orchestrating Multi-Agent Multi-Step Reasoning via Graph Memory with Reinforcement Learning

• Authors: Zhongyu Wang
• URL: https://arxiv.org/abs/2609.14066
• Abstract:

  Although existing multi-agent Retrieval-Augmented Generation (RAG) systems have demonstrated promise on complex multimodal reasoning tasks, they remain fundamentally limited in reasoning depth and memory structure, suffering from inadequate retrieval and state blindness when answering knowledge-intensive questions. To address these limitations, we propose GraMRAG, a graph memory-guided multi-agent RAG framework that integrates a dynamic multimodal memory graph to enable stable, multi-step multimodal reasoning. We introduce a vision-text bridged reasoning paradigm that unifies multi-scale entity cropping with a ReAct-style visual toolchain, enhancing the long-horizon cross-modal reasoning capability. We further construct a multimodal memory graph that formalizes agent reasoning as a dynamic directed acyclic graph (DAG), explicitly modeling action-observation dependencies to mitigate state blindness and suppress redundant retrieval. Moreover, we propose Topology-Aware Policy Optimization (TAPO) that leverages graph topology for critical path identification and targeted node pruning, enabling fine-grained credit assignment across multi-step reasoning trajectories. Extensive experiments on challenging multimodal benchmarks demonstrate that our approach consistently outperforms existing baselines and achieves state-of-the-art performance on complex long-horizon reasoning tasks.

---

321. AGENTQ: Quantization-Conditioned Backdoor Attacks on LLM Agents

• Authors: Xiaoqun Liu , Qiben Yan
• URL: https://arxiv.org/abs/2609.14060
• Abstract:

  Quantization is one of the default deployment paths for open-weight LLM agents, but it is not behavior-preserving: an adversary can release a full-precision checkpoint that passes audits yet misbehaves once quantized, termed as quantization-conditioned attack (QCA). Prior QCA work targets free-text generation, where harm is mediated by a human reader. In contrast, the agentic setting poses a more severe risk: the triggered payload is a structured function that can be executed without human oversight. We present the first study of QCA against LLM agents. We find that directly adapting prior backdoor-injection methods can produce malicious behavior after quantization, but substantially degrades benign utility, rendering the resulting attacks impractical. To understand the true upper bound of the threat, we propose AGENTQ, an attack framework that combines layer-banded LoRA injection with partial-PGD repair over a multi-codebook quantization-equivalence class. AGENTQ preserves normal agentic capability while concentrating malicious behavior in the quantized model. Across three trigger-action pairs and three codebooks (NF4, FP4, INT8), AGENTQ reaches up to 100% post-quantization attack success rate with minimal loss of benign utility, underscoring the need to make quantization-aware safety evaluation a standard requirement before open-weight agents are deployed.

---

322. Confuse the Model, Control the Flow: Understanding and Mitigating Privacy Leakage from LLM Agents with Information Flow Control

• Authors: Minsun Shim , Ramisha Raida Karim , Ruthwik Jakkula , Kaiwen Zhou , Xin Liu , Xin Eric Wang , Zhou Li
• URL: https://arxiv.org/abs/2609.14003
• Abstract:

  Personal AI agents built on large language models (LLMs) are increasingly given access to a user’s private data and communications in order to provide personalized assistance. This access creates a persistent privacy risk: the agent must decide whether a given sensitive information should be disclosed to a particular party. Existing defenses address this by making the agent’s backend LLM more privacy-preserving through stronger system prompts, training, or explicit consent-checking procedures, but this approach has a structural challenge: whenever enforcement is a judgment the LLM makes over the same conversational context an adversary controls, the enforcement mechanism and the attack surface coincide. We demonstrate this against existing defenses with three new attacks that require only ordinary agent interaction and no prompt injection: Collaborative Workspace Lure reframes an extraction attempt as collaborative work; Semantic Obfuscation Attack induces disclosure through omission rather than through anything the agent writes; and Channel Decoupling Attack splits the extraction request and the disclosure across independent channels. All three achieve substantially higher leak rates than the attacks these defenses were originally designed to withstand. Guided by this observation, we present FLOWSEAL, a defense that enforces confidentiality through a tool-level interceptor outside the LLM’s context, grounded in data provenance and an information-flow-control lattice with controlled declassification. Evaluated across three benchmarks, five prompt-based baselines, and eight attacks, including a real agent executing live tool calls through MCP, FLOWSEAL reduces leak rates to near zero (e.g., 52.2% to 0.5% against Collaborative Workspace Lure) while preserving task utility, regardless of the underlying LLM backend.

---

323. Rethinking the Implications of Human Feedback for Preference Learning in Human-Robot Collaboration

• Authors: Qiping Zhang , Kate Candon , Debasmita Ghose , Marynel Vázquez
• URL: https://arxiv.org/abs/2609.13982
• Abstract:

  In Human-Robot Interaction, the standard approach to learn a reward model that represents human preferences for robot behavior consists of three steps. First, the robot collects limited direct evidence from human feedback (e.g., positive or negative binary feedback). Then, the robot utilizes the direct evidence to derive accepted or rejected labels to feasible but unchosen actions using fixed implication rules. Finally, the robot updates the reward model with both the direct and derived evidence. Unfortunately, the fixed rule can hinder preference learning: in a user study with two collaborative simulation environments, human-provided implication labels often differed from the standard fixed rule, and using the human labels substantially improved reward learning with the Preference Learning from Implicit and Explicit Feedback (PIE) algorithm. Consequently, we propose IMPLIED, an implication modeling method that treats fixed-rule implications as an initial guide while learning to infer and revise accepted and rejected action labels over time. Across evaluations on recorded human-robot interaction trajectories and a physical robot pizza-making study, IMPLIED predicts human implications more accurately than the fixed rule approach and LLM baselines, approaching the performance of a human-label oracle. In turn, IMPLIED reduces preference-estimation error and leads to robot actions that are more often rational with respect to a combined reward (which includes the true preference reward and a task-specific reward) compared to baselines. By learning to reason about the implications of human feedback, this work enables more faithful and efficient robot behavior adaptation during human-robot collaboration.

---

324. Mizan: A National Benchmark for Evaluating Large Language Models on Iraqi Arabic and the Iraqi Civic Context

• Authors: Nawar S. Alseelawi , Mustafa S. Aljumaily
• URL: https://arxiv.org/abs/2609.13980
• Abstract:

  Arabic large-language-model (LLM) evaluation has matured around Modern Standard Arabic (MSA): aggregated leaderboards such as the Open Arabic LLM Leaderboard (OALL), HELM Arabic, and BALSAM rank models across dozens of MSA tasks, and frontier systems increasingly saturate them. Dialectal Arabic, the language Iraqis actually speak, remains nearly invisible to this infrastructure. We introduce Mizan (“the balance”), Iraq’s national benchmark for evaluating LLMs on Iraqi Arabic and the Iraqi civic context: an MSA baseline track paired with an Iraqi track across six axes (dialect comprehension, dialect generation, bidirectional MSA-Iraqi translation, Iraq-specific knowledge, official-document field extraction, and safety), built from 340 originally authored, dually reviewed items with statistically audited answer positions and Wilson intervals on every published score. A pilot evaluation of 27 systems, spanning closed frontier models three days after release, open weights across size tiers, and an Arabic trio of commercial, open-specialized, and sovereign systems, yields four findings. The MSA track saturates while the Iraqi track discriminates, with a consistent 14-18-point per-model gap and statistically tied leaders. Official-document extraction confines every system to 32-56. Arabic-focused specialization behaves as MSA specialization: two dedicated Arabic models score below a size-matched generalist on the Iraqi track. And the safety-hardened tier of the newest model family deterministically refuses innocuous dialect-comprehension items as policy violations, an over-refusal mode invisible to MSA benchmarks. The platform enforces an integrity protocol of immutable snapshots, verification certificates, a human publication gate, and public retraction, all exercised during this study. Code and the public development set accompany the paper.

---

325. SGWIB:Sliced Gromov-Wasserstein Information Bottleneck for Video Highlight Detection

• Authors: Hanjuan Huang , Yung-Chieh Yeh , Hsing-Kuo Pao
• URL: https://arxiv.org/abs/2609.13966
• Abstract:

  Video highlight detection aims to identify temporally important segments that capture the most informative or engaging events in a video. Reliable prediction therefore requires not only discriminative segment representations but also preservation of the temporal relationships among neighboring and distant segments. The information bottleneck principle has proven effective for learning compact and task-relevant representations, yet it has not been explored for video highlight detection, and applying conventional formulations directly would overlook inter-segment relational structure and distort highlight relevant temporal organization during compression. We therefore introduce the Sliced Gromov-Monge Gap (SGMG), a structure aware regularizer that measures the excess relational distortion induced by a prescribed source-to-bottleneck mapping relative to an optimal sliced structural correspondence. Building on SGMG, we develop SGWIB, an information-bottleneck framework for single-modal video highlight detection that learns compact bottleneck representations while preserving inter-segment temporal structure. We further introduce Home-Away-Related Contextual Pseudo-Labels and a contextual disentanglement module that reduce sports-specific contextual bias by separating highlight oriented information from contextual patterns. Experiments on MrHiSum and MoSu show that SGWIB attains the best Kendall’s tau, Spearman’s rho, mAP@50, and mAP@30 among the compared single-modal methods on both datasets. On MrHiSum, the visual model improves the strongest previous results by 0.031, 0.031, 0.87, and 0.75 on these four metrics, respectively. These results show that structure-aware information-bottleneck regularization combined with contextual disentanglement improves segment-level highlight prediction.

---

326. Thought without systematicity? Evaluating reasoning models on rule induction tasks

• Authors: Simon Schug , Brenden M. Lake
• URL: https://arxiv.org/abs/2609.13948
• Abstract:

  A central tenet of human cognition is systematicity, the principle that understanding one concept is inherently tied to understanding close variations of that concept. Do reasoning models robustly exhibit such systematicity? If so, we would expect consistent performance on structurally equivalent variants of the same task. Here, we extend established rule induction tasks from cognitive science to assess the systematicity of thought in current reasoning models. Each task family has compositional structure that we use to create structurally equivalent task variations through task isomorphisms such as recombination and substitution. We find that despite being able to correctly solve a task, models often fail on structurally equivalent variants of the same task. These findings suggest that many model behaviors lack systematicity, rendering it difficult to robustly establish the cognitive abilities of reasoning models beyond the particular contexts they were evaluated in.

---

327. Hardware-Aware Learned Representation Compression for Distributed In-Sensor Vision

• Authors: Chengwei Zhou , Abu Masum , Xuming Chen , Mehran Moghadam , Sreetama Sarkar , Arnab Sanyal , Md Abdullah-Al Kaiser , M. Hassan Najafi , Sercan Aygun , Gourav Datta
• URL: https://arxiv.org/abs/2609.13947
• Abstract:

  In-sensor computing reduces the cost of transmitting high-resolution image data by performing early-stage processing near the sensor. However, the logic chip integrated with a CMOS image sensor (CIS) is tightly constrained in compute and memory, limiting conventional deep neural network partitioning. We present OASIS, a distributed in-sensor vision framework that uses a lightweight encoder to generate compact, task-relevant representations before off-chip transmission. The encoder is trained end-to-end using task, entropy, and reconstruction objectives, while the decoder is used only during training. OASIS supports two complementary deployment paths. The first applies 4-bit quantization and Huffman coding while preserving the spatial structure required by classification and dense-prediction tasks. The second uses Sobol-based hyperdimensional computing (HDC) to transform the encoder latent into a fixed-dimensional binary hypervector for associative-memory classification. For the SwinViT-based VWW model, mapping a $3\times3\times8$ latent to a 64-dimensional hypervector provides an additional $1.77\times$ communication reduction with less than one percentage point of accuracy loss relative to the 128-dimensional configuration, yielding an overall $18{,}816\times$ reduction compared with raw 8-bit image transmission. We implement the digital near-sensor pipeline on an AMD Xilinx Zynq UltraScale+ FPGA and characterize it using direct board-level power measurements and Vivado post-implementation analysis, together with circuit-simulated CIS models and a 7-nm ASIC projection. Across visual wake-word classification, hand tracking, and eye tracking, OASIS reduces total system energy by approximately $2\times$-$4.5\times$ while maintaining competitive accuracy, demonstrating a practical hardware-algorithm co-design path for communication-efficient in-sensor vision.

---

328. CRITICS - Critical Science Without Borders: Language Models to Promote Critical Thinking in Science Education

• Authors: Rodrigo Agerri , Itziar Aldabe , Elena Cabrio , Mark Cieliebak , Jan Deriu , Mariana Flores , Jurgita Kapociute-Dzikiene , Dovile Kuiziniene , Arantza Rico , Aritz Ruiz-Gonzalez , Aitor Soroa , Mantas Vaskevicius , Serena Villata
• URL: https://arxiv.org/abs/2609.13942
• Abstract:

  The CRITICS project addresses science accessibility and literacy by converging advanced Machine Translation (MT) based on Large Language Models (LLMs) with educational technology. By leveraging MT systems specifically optimized for scientific content, educational institutions can provide accurate, culturally relevant translations of scientific materials in students’ native languages, ensuring that complex scientific concepts are comprehensible while maintaining technical accuracy. Building on these translations, the project explores the design and evaluation of innovative science teaching-learning proposals grounded in curriculum-aligned teaching-learning. Thus, CRITICS will investigate key components of scientific argumentation and critical thinking practices together with textual feedback aligned with learning objectives and assessment criteria inspired by competence-based evaluation frameworks. CRITICS aims to break down language barriers to accessing cutting-edge research and educational materials currently available only in high-resourced languages, thereby facilitating the democratization of scientific knowledge and fostering critical thinking in science education.

---

329. Finite-Time Node Separation in Recurrent Graph Neural Networks with Persistent Gaussian Perturbations

• Authors: Mostafa Haghir Chehreghani
• URL: https://arxiv.org/abs/2609.13920
• Abstract:

  Persistent Gaussian perturbations have been shown to prevent asymptotic oversmoothing in recurrent Graph Neural Networks (GNNs) by ensuring a positive stationary Dirichlet energy. However, this global energy bound does not guarantee that individual node representations remain distinct at finite depths. In this paper, we provide a complementary finite-time analysis of the same persistent-noise architecture. Let (d) denote the representation dimension and (\sigma) the noise standard deviation. We first prove an exact second-moment decomposition for the expected squared distance between any two node representations, yielding the universal lower bound (2\sigma^2 d) at every positive time step without contraction or stationarity assumptions. More precisely, conditional pairwise distances have a noncentral chi-square characterization: the noncentrality parameter is the deterministic message-passing separation normalized by (2\sigma^2). This yields dynamics-aware fixed-time and finite-horizon near-collision bounds that retain information discarded by the central worst-case analysis. The earlier central Gaussian bound is recovered as the worst-case zero-separation case. We additionally prove almost-sure pairwise noncollision, derive a uniform finite-horizon guarantee, and establish permutation equivariance in distribution for the stochastic dynamics and permutation-invariant graph outputs. Our results complement the asymptotic energy analysis of prior work and provide rigorous finite-time guarantees on node-level representation separation.

---

330. DiTAR+: Dual Optimization for Robust Autoregressive Diffusion Speech Synthesis

• Authors: Ziyu Zhang , Tianlun Zuo , Hanzhao Li , Haoyu Zhang , Lei Xie
• URL: https://arxiv.org/abs/2609.13909
• Abstract:

  Continuous-latent Autoregressive Diffusion Transformer (AR-DiT) models have demonstrated immense potential in zero-shot speech generation. However, they still suffer from limited decoding stability when synthesizing long utterances or complex linguistic structures. This instability primarily stems from a restricted historical receptive field and an acoustic inertia dependency within the diffusion decoder, which causes the model to ignore semantic conditions. To address these challenges, we propose DiTAR+, a dual-optimization framework. First, we introduce Dilated Context Sampling to expand the macro-level historical receptive field without violating physical temporal continuity, thereby preventing cumulative error propagation. Second, we propose Hierarchical Acoustic Masking to prevent shallow layers from attending to acoustic pre-context, explicitly decoupling semantic alignment from acoustic detail reconstruction. Extensive experiments show that our framework effectively mitigates pronunciation errors and semantic hallucinations, enhances generation robustness on challenging sentences, and maintains exceptionally high speaker similarity throughout the entirety of long-form utterances. On the linguistically challenging ZH-Hard set, DiTAR+ reduces the word error rate from 12.478% to 9.893%, and on extended utterances of 25 to 35 seconds it improves speaker similarity from 0.741 to 0.759 while simultaneously lowering the word error rate from 2.778% to 2.173%, outperforming both discrete-token and pure flow-matching baselines.

---

331. Phorecaster365: A Human-Supervised Reference Architecture for Hybrid Pharmaceutical Sales Forecasting and Planning Decision Support

• Authors: Houman Kazemzadeh , Kamyar Naderi
• URL: https://arxiv.org/abs/2609.13907
• Abstract:

  Pharmaceutical sales forecasts inform planning across products, regions, and distribution channels, yet their interpretation depends on inventory availability, transaction semantics, product lifecycle, and the information available when each forecast is issued. A model prediction alone does not preserve these conditions or establish whether a forecast is suitable for operational use. We present Phorecaster365, a human-supervised reference architecture that connects enterprise resource planning data to reviewable pharmaceutical sales forecasts. The architecture separates source ingestion, product-region series construction, temporally valid feature generation, statistical and machine-learning modeling, ensemble formation, uncertainty assessment, planner review, and lifecycle governance. Its central intermediate representation is a forecast context package that preserves the source snapshot, forecast target, temporal cutoff, available covariates, data-quality state, and hierarchy version. A corresponding forecast evidence package links predictions to model and calibration versions, exceptions, human adjustments, and publication history. Development experience with a synthetic panel of 10,950 daily records across 30 product-region series informs the design. Historical experiment summaries are retained only as descriptive evidence of development because their evaluation does not establish independent predictive validity. We specify a rolling-origin evaluation protocol, baseline and ablation requirements, uncertainty and robustness assessments, and a staged pathway from synthetic testing to use in governed planning. The contribution is an implementation-neutral system design and validation framework; the report does not establish real-world forecasting accuracy, comparative superiority, or operational benefit.

---

332. When Malicious Instructions Persist: Persistent Memory Poisoning Attack on Harness-Based Agents

• Authors: Shuhuai Huang , Jingfeng Zhang , Hong Jia
• URL: https://arxiv.org/abs/2609.13889
• Abstract:

  Harness design has transformed the development of LLM-based agents by integrating memory, tool use, and runtime control. However, this design also introduces security and privacy risks because malicious instructions from external sources may be written into persistent memory and persist across sessions. To study this risk, we propose PMPA, a Persistent Memory Poisoning Attack against harness-based agents. PMPA embeds malicious instructions into benign external sources and induces the victim agent to write them into persistent memory without directly accessing to the agent framework. Once stored, the poisoned memory can be retrieved in later sessions, triggering additional malicious actions and causing privacy leakage. We evaluate PMPA on OpenClaw and Claude Code across different backbone LLMs, input modalities, and trigger scenarios. Across all settings, PMPA achieves average Injection Success Rate (ISR) and Cross-session Attack Success Rate (C-ASR) of 73.7%/ 55.5% on OpenClaw and 66.9%/ 81.7% on Claude Code, while preserving benign task performance on both systems. We further evaluate a targeted prompt-level defense and find that it can reduce memory injection in many settings, but provides limited protection once the persistent memory has been poisoned.

---

333. Trustworthy, Explainable, and Sustainable Decentralized Intelligence for 6G Networks

• Authors: Giovanni Perin , Michele Rossi , Enrique Tomás Martínez Beltrán , Fernando Torres-Vega , José María Jorquera Valero , Manuel Gil Pérez , Eunjeong Jeong , Nikolaos Pappas , Farah Abed Zadeh , Chamara Sandeepa , Bartlomiej Siniarski , Madhusanka Liyanage , Betül Güvenç Paltun , Leyli Karaçay , Ioannis Pitsiorlas , Marios Kountouris
• URL: https://arxiv.org/abs/2609.13872
• Abstract:

  As 6G networks transition from theoretical frameworks into operational realities, artificial intelligence (AI) evolves from an add-on optimization tool into a distributed and interconnected structural layer. Unlike previous network generations that mostly relied on centralized cloud analytics platforms, AI-native 6G networks operate across a dynamic, multi-domain edge-cloud continuum where data originates from heterogeneous sources including user devices, radio access networks, sensing infrastructures, and vertical applications. Centralizing this massive volume of data creates severe communication overhead, unacceptable latency bottlenecks, single points of failure, and complex cross-domain governance challenges. Consequently, decentralization becomes a fundamental architectural requirement for future 6G network intelligence and zero-touch operations. Security serves as the primary enabler of this decentralized paradigm. Critical security functions, such as real-time threat detection, physical-layer attack mitigation, slice protection, and intrusion detection, require immediate access to local context and telemetry before operational data loses its value. However, moving intelligence to the edge via collaborative paradigms like federated learning (FL) and decentralized FL (DFL) introduces complex trade-offs. System security cannot be addressed in isolation; it is deeply intertwined with equally important aspects like trustworthiness, explainability, and energy sustainability. Taking these aspects into account, this paper develops a unified perspective on decentralized intelligence for 6G, arguing that decentralization, trustworthiness, explainability, and sustainability must be designed jointly rather than treated as independent requirements

---

334. Bangla Sentence Function Classification: Corpus Development, Model Benchmarking, and Interpretability

• Authors: Swapnil Kundu Argha , Abdullah Al Shafi , Rowzatul Zannat , Shoumik Barman Polok , Abdul Muntakim , Jannatul Ferdousi , M.A. Moyeen
• URL: https://arxiv.org/abs/2609.13869
• Abstract:

  Automatic sentence function identification is important for many downstream natural language processing (NLP) applications such as dialogue systems, text-to-speech synthesis, and machine translation. However, benchmark resources for Bangla sentence function classification remain limited. To mitigate this gap, this paper introduces a corpus of 10,000 Bangla sentences, manually annotated into four functional categories, namely declarative, interrogative, imperative, and exclamatory. The corpus is nearly balanced across the four classes, with high annotation reliability reflected by a Fleiss' Kappa of 0.82. Furthermore, we evaluate multiple feature representations, including Bag-of-Words (BoW), TF-IDF, and Word2Vec, with several classical machine learning classifiers. In addition, two heterogeneous ensemble models, namely Single-Level Ensemble (SLE) and Double-Level Ensemble (DLE), are utilized to improve classification performance. Experimental results show that TF-IDF consistently outperforms Word2Vec, likely due to its ability to emphasize discriminative lexical cues associated with sentence functions, particularly given the relatively small corpus used to train Word2Vec. The DLE model with TF-IDF features achieves the best performance with accuracy and macro-F1 of 0.95, demonstrating the effectiveness of sparse lexical representations and heterogeneous ensemble learning for this task. Further cross-validation confirms the robustness of the approach, while LIME-based interpretability provides insights into model predictions. The developed corpus and model benchmarking establish strong baselines for Bangla sentence function classification.

---

335. ReWeight: Leveraging Human Data for VLA Post-Training via Demonstration Retrieval and Sample Weighting

• Authors: Chenwei Wang , Dianye Huang , Match W.L. Ko , Chenjia Bai , Zhongliang Jiang
• URL: https://arxiv.org/abs/2609.13851
• Abstract:

  Post-training vision-language-action (VLA) models for specific robots and tasks requires in-domain demonstrations, yet collecting diverse robot data is costly. Egocentric human demonstrations provide a scalable alternative, but directly mixing human and robot data can introduce cross-embodiment discrepancies and degrade policy performance. To address this challenge, we introduce ReWeight, a framework that incorporates human data into VLA post-training through demonstration-level retrieval and sample-level weighting. ReWeight learns a cross-embodiment visuomotor representation that combines visual observations with future actions to measure behavioral similarity between human and robot demonstrations. Based on optimal transport, it retrieves human demonstrations relevant to the target robot data and assigns larger weights to samples with smaller cross-embodiment discrepancies. We evaluate ReWeight using $\pi_{0.5}$ across eight simulation tasks and four real-world tasks under both clean and randomized settings. In simulation, ReWeight improves the average success rate of post-trained $\pi_{0.5}$ from 39% with only robot data and 44% with randomly mixed human-robot data to 57%. In the physical experimental setting, it achieves an average success rate of 68.8%, outperforming the baselines by 28.8% and 13.8%, respectively. Overall, ReWeight provides an effective paradigm for transforming abundant egocentric human experience into transferable supervision for robot learning. (Project webpage: this https URL )

---

336. LePlanner: An Iterative Amortized Controller For World Models

• Authors: Saksham Bansal , Om Naphade , Chayan Aggarwal , Vrishin M
• URL: https://arxiv.org/abs/2609.13845
• Abstract:

  World models trained with joint-embedding predictive architectures learn compact, structured latent representations from physical interaction, yet planning in these latent spaces typically relies on one of two costly approaches. Search-based planners such as CEM, MPPI, and iCEM optimize action sequences through many predictor rollouts, achieving strong performance at the cost of high per-decision compute and latency. Policy-based methods amortize inference into a single forward pass but can degrade on contact-rich tasks where the demonstration distribution is multimodal. We propose LePlanner, an amortized iterative controller that learns to construct and refine latent action sequences through a frozen world-model predictor. LePlanner is trained with an arrival-and-hold objective that encourages the controller to reach the goal at the earliest feasible horizon and remain there. This addresses horizon-reset procrastination, a failure mode in which repeated receding-horizon replanning continually postpones goal arrival. An additional action-Gaussian loss keeps generated actions near the support of the offline dataset. Across navigation, contact-rich manipulation, and continuous-control environments, LePlanner matches or exceeds search-based planners while requiring an order of magnitude fewer predictor evaluations and 3-49x lower wall-clock time per decision. It achieves success rates of 98% on PushT, 100% on Reacher, 100% on TwoRooms, and 92% on the OGBench Cube task. These results show that much of the structure discovered through online search can be amortized into a lightweight iterative policy, enabling fast, horizon-aware, nonlinear physical control without online optimization.

---

337. CRAF: Cross-View Residual-Aware Fusion for Deepfake Speech Detection

• Authors: Minh-Xuan Phan , Khalid Zaman , Candy Olivia Mawalim , Masashi Unoki
• URL: https://arxiv.org/abs/2609.13842
• Abstract:

  Recent advances in speech synthesis and voice conversion have made deepfake speech increasingly realistic, making generalization to unseen spoofing attacks a critical challenge. Pretrained speech and audio models offer a promising direction for improving robustness to such unseen attacks. Self-supervised learning (SSL) models capture fine-grained, low-level acoustic characteristics, whereas Auditory Large Language Models (ALLMs) provide higher-level contextual representations. These complementary views can provide useful cues for improving generalization to unseen attacks. However, direct fusion does not explicitly disentangle information shared across the two views from view-specific complementary information, limiting effective cross-view integration. To address this, we propose CRAF, a cross-view residual-aware fusion framework that uses ALLM-guided cross-view attention to enrich SSL representations and adopts ALLM as a high-level reference to separate ALLM-explainable information from complementary SSL residual information. The residual is selectively refined through adaptive gating and integrated through SSL-primary fusion. Experiments on ASVspoof 5 show that CRAF with Kimi-Audio achieves an EER of 5.96% and a minDCF of 0.1192, demonstrating robustness to unseen spoofing attacks.

---

338. Exploring Automated Vulnerability Identification in JavaScript Code Using Large Language Models

• Authors: Manit Kaushik , Ishir Bhardwaj , Pranav Gupta , Pankaj Jalote , Arun Balaji Buduru
• URL: https://arxiv.org/abs/2609.13816
• Abstract:

  JavaScript powers approximately 98.8% of all websites, making vulnerabilities in its code a significant security risk, yet existing detection approaches such as Static Application Security Testing (SAST) tools often fail to identify many real-world vulnerabilities when applied to isolated code snippets. This paper presents an empirical study of Large Language Model (LLM)-based vulnerability identification for JavaScript programs, evaluating three LLM families (Gemini 1.5 Flash, GPT-4o Mini, DeepSeek-R1-Distill-Llama-8B) across multiple prompting strategies (zero-shot, chain-of-thought, few-shot) and fine-tuning approaches on a dataset of 1,125 JavaScript code snippets spanning five Common Weakness Enumeration (CWE) categories: Injection (CWE-74), OS Command Injection (CWE-78), Cross-Site Scripting (CWE-79), SQL Injection (CWE-89), and Uncontrolled Resource Consumption (CWE-400). Our experiments show that LLMs substantially outperform traditional SAST tools on snippet-level vulnerability identification, with a fine-tuned Gemini 1.5 Flash model achieving 60% detection accuracy compared to near-zero performance from rule-based analyzers. We find that fine-tuning improves accuracy from 29% to 60%, Chain-of-Thought prompting benefits reasoning-capable models such as GPT-4o Mini, few-shot prompting is effective for polymorphic vulnerabilities such as Cross-Site Scripting, and performance varies across vulnerability categories, reaching up to 84% accuracy for structured vulnerabilities such as SQL Injection. These results indicate that LLMs provide a practical approach for automated vulnerability identification in JavaScript code, particularly when combined with task-aligned supervision, though they should complement rather than replace existing security analysis workflows due to limited recall and uneven performance across vulnerability types.

---

339. Understanding the Limits of Agentic ICD Coding

• Authors: Chong Yock Eng , Yushi Cao , Yiming Chen , Kezhi Mao , Hongchao Jiang
• URL: https://arxiv.org/abs/2609.13806
• Abstract:

  ICD-10-CM codes are alphanumeric codes used in the US to classify diagnoses and injuries for medical billing and epidemiological reporting. Standard ICD-10-CM benchmarks report aggregate metrics that obscure performance on complex coding scenarios. We evaluate neural, workflow, and agentic systems on a rarity-stratified set of MIMIC-IV discharge summaries and identify two orthogonal failure modes. Neural classifiers exhibit a 0.43 micro-F1 gap between rare and common codes. Workflow systems handle rare codes well but score near zero on injury and external cause codes that require multi-step guideline following. A tool-augmented agentic configuration with structured access to official ICD-10-CM reference materials recovers up to 0.34 micro-F1 on this subset. No single system dominates across all conditions.

---

340. GEAR: From Dynamic Encoding to Dynamic Activation in Social Trajectory Prediction

• Authors: Jiaheng Chen , Jiaxing Li , Leixia Wang , Jianan Ju , Tinghe Zhang
• URL: https://arxiv.org/abs/2609.13778
• Abstract:

  Human trajectory prediction requires modeling both individual motion patterns and social interactions among agents. Existing methods have made substantial progress by using attention mechanisms, graph structures, and temporal encoders to capture dynamic social context. However, most of them primarily focus on how social information is encoded, while paying less explicit attention to how the encoded social context should take effect during future trajectory generation. In this paper, we argue that dynamic social encoding does not necessarily imply dynamic social activation. The same interaction context may require different activation strengths across future horizons and scene densities: social cues should be strengthened when interaction evidence is strong, but suppressed when they are weak or noisy. To address this issue, we propose GEAR, a generation-aware bias activation model for human trajectory prediction. Built upon a bias-decomposed trajectory generation formulation, GEAR dynamically activates the individual-motion and social-resonance bias terms at each future step before final trajectory composition. This allows the model to explicitly control when and how strongly individual and social bias components participate in generation. Experiments on ETH-UCY, SDD, and NBA show that GEAR consistently improves the resonance-based baseline and achieves competitive state-of-the-art performance. Further analyses of activation patterns and density-grouped errors validate the importance of calibrating encoded social context during trajectory generation. Our code is available at this https URL .

---

341. On the Equivalence of Stochastic Control and Path Space Formulations for Schrödinger Bridges over Compact Connected Lie Groups

• Authors: Hamza Mahmood , Georgiy A. Bondar , Abhishek Halder , Adeel Akhtar
• URL: https://arxiv.org/abs/2609.13758
• Abstract:

  We establish the equivalence between the stochastic optimal control and path space formulations of the Schrödinger bridge problem (SBP) for the kinematic equation on a compact connected Lie group. Using the geometric concepts of horizontal lift and stochastic anti-development, we derive a Girsanov-type change-of-measure result, and show that the expected control energy equals the relative entropy of the controlled path law with respect to the reference Wiener measure. Thus, the SBP is equivalently a path space relative entropy minimization problem subject to prescribed endpoint this http URL result has three useful implications. From an analytic viewpoint, the shown equivalence helps prove the existence and uniqueness of the SB. From a probabilistic viewpoint, it helps interpret the SB as the most probable deviation of the uncontrolled stochastic dynamics consistent with the endpoint constraints. From a computational viewpoint, it allows using static Sinkhorn recursions to directly solve the relative entropy minimization problem and compute the optimal path measure. We illustrate the equivalence numerically on the torus $\mathbb{T}^2$. The code is publicly available at: this https URL

---

342. HarnessBandit: Joint Learnability-Transferability Scheduling for Multi-Harness Agentic Reinforcement Learning

• Authors: Hongliang Wei (1 and 2), Xiaobing Tu (2), Yinggui Wang (2), Zhengxi Liu (2), Rongkun Xue (2), Jinkui Ren (2), Xiantao Zhang (2), Debin Zhao (1), Xiaopeng Fan (1) ((1) Harbin Institute of Technology, (2) Alibaba Cloud)
• URL: https://arxiv.org/abs/2609.13739
• Abstract:

  Language-model agents are increasingly deployed through diverse harnesses that differ in system prompts, tool schemas, control loops, and trajectory formats. The same model can perform unevenly across these interfaces, making robustness to harness variation an important objective. A natural approach is to train a shared policy through multiple harnesses, but doing so introduces a scheduling problem: each training step should favor a harness that currently provides a useful learning signal while also producing an update that benefits the other harnesses. We develop HarnessBandit, an online scheduler that selects one harness per optimizer step. After a group-relative policy optimization (GRPO) update, it observes learnability – the mean absolute advantage on the batch – and transferability – the cosine between a low-dimensional gradient sketch of the current harness and exponential moving averages of the remaining harnesses. The two signals are fused after pooled sliding-window min-max normalization and sampled with a visit-dependent bonus and an explicit exploration floor. We train Qwen3.5-2B across six harnesses on ClawGym and evaluate on PinchBench (held-out tasks, in-distribution OpenClaw) and ClawEval (held-out tasks and harness). HarnessBandit improves over mixed-batch multi-harness training on both benchmarks, while training diagnostics indicate that learnability and transferability provide distinct, evolving signals.

---

343. PolicyMem: Geometric Policy Memory for LLM Governance

• Authors: Yuanchen Bei , Zhengzhang Chen , Yanjun Zhao , Haoyu Wang , Hanghang Tong , Haifeng Chen
• URL: https://arxiv.org/abs/2609.13734
• Abstract:

  As large language models (LLMs) are increasingly deployed in real-world high-stakes applications, effective governance has become essential. Existing safeguards largely follow two paradigms: learning-based guards provide strong semantic discrimination but couple policy behavior to trained models and taxonomies, while programmable frameworks offer flexible control but require substantial manual prompt and workflow engineering. Neither externalizes policies as reusable operational states, making it difficult to consistently reuse policy evidence across detection, intervention, and verification. In this paper, we introduce PolicyMem, a geometric policy memory that externalizes natural-language policies as reusable geometric memory objects represented by low-rank subspaces in a shared representation space. A memory writer compiles natural-language policies into policy memory slots, and query-response pairs read the policy memory through projection energy. The resulting policy-evidence profile directly mediates the safety verdict and is reused for policy attribution and post-intervention verification. Coupled with a response rewriter, PolicyMem enables a detect-rewrite-verify loop for LLM governance. Across five widely used benchmarks, PolicyMem achieves state-of-the-art unsafe behavior detection while enabling effective policy attribution, rewriting, and post-intervention verification through the shared policy memory.

---

344. TyPatch: Transforming Patches into Typestate Rules for Kernel Bug Detection

• Authors: Ruoyu Wang , Tuo Li , Jia Li
• URL: https://arxiv.org/abs/2609.13728
• Abstract:

  Historical Linux kernel patches capture defect knowledge that applies beyond their original repair sites. Recent work has shown that large language models (LLMs) can generate static-analysis checkers from historical patches and use them to uncover new kernel bugs. However, complete-checker generation requires the model both to recover the defect semantics expressed by a patch and to implement sophisticated program-analysis machinery, including object tracking, alias analysis, path-state maintenance, and interprocedural propagation. Coupling these responsibilities in a single end-to-end code-generation task can turn a simple defect rule into an unstable and expensive analyzer-implementation problem. To address this problem, we present TyPatch, which decouples patch-specific defect semantics from analyzer implementation. An LLM translates each patch into a typestate rule specifying its tracked object, actions, guards, transitions, and violations. A shared backend then executes these rules, binding their actions to program events, tracking object identity across aliases, propagating typestate along program paths, and producing reports for all rules. On Linux v6.16, TyPatch finds 559 distinct bugs, 121 of which have been confirmed by kernel developers. In a matched 38-patch comparison with the state-of-the-art complete-checker construction workflow, TyPatch uses 88.3-90.1% fewer generation tokens, while its initial report pools achieve 3.42-14.95$\times$ the precision of those produced by that workflow.

---

345. Oops, Not Now: PEARL, a RAG-Based Support Agent for Gameplay and What Players Want from AI Help

• Authors: Jiahong Li , Sai Siddartha Maram , Atieh Kashani , Ulia Zaman , Zhiyu Lin , Cameron Marano , Roger Azevedo , Jichen Zhu , Magy Seif El-Nasr
• URL: https://arxiv.org/abs/2609.13718
• Abstract:

  AI-powered gameplay support agents hold promise for game-based learning, yet grounding generative models in structured game data remains an open challenge. We present PEARL (Parallel Education Agent for Reflection and Learning), a dual-component Retrieval-Augmented Generation (RAG) system that combines semantic knowledge retrieval with structural board-state matching to deliver contextualized scaffolding in Parallel, a puzzle game for learning parallel programming. PEARL operates on two input streams (natural language queries and board topology), retrieving both conceptual explanations of gameplay moves and peer-generated board states as evidence: capabilities unavailable to a standard Large Language Model (LLM) with game state access alone. In a qualitative evaluation (N=10) comparing PEARL against an existing community-based Open Player Model (OPM) visualization system, participants preferred the visualization system on perceived usefulness and reported higher frustration with PEARL; five of ten minimized or abandoned the AI tool during play. Proactive delivery, generic responses, and trust deficits drove disengagement, while a subset of four participants found PEARL’s grounded explanations complementary to visualization in specific contexts where they initiated the interaction. We position PEARL as a deployed design probe whose failure modes inform a concrete design agenda for AI gameplay support, captured as seven open problems for the community.

---

346. Gap Entropy and Almost Instance-Wise Optimal Best-Arm Identification

• Authors: Jiarui Yao , Jiaxi Zhao , Xiangxin Zhou
• URL: https://arxiv.org/abs/2609.13703
• Abstract:

  In the best-arm identification problem, we are given $n$ stochastic arms with unknown means and wish to identify the arm with the largest mean with probability at least $1-\delta$, using as few samples as possible. We consider independent Gaussian rewards with unit variance and means in $[0,1]$. Chen and Li [2016] conjectured that the instance-wise sample complexity of this problem is characterized by the gap entropy, up to an additive term arising from the two-arm problem. In this paper, we resolve their gap-entropy and almost instance-wise optimality conjectures. For an instance $I$, let $\Delta_{[i]}$ be the gap between the largest and the $i$-th largest mean, let $H(I)=\sum_{i=2}^{n}\Delta_{[i]}^{-2}$, and let Ent$(I)$ denote the entropy of the normalized complexities of its dyadic gap groups. For every $0<\delta<0.1$, we show that the order-oblivious instance-wise lower bound is $ {\Theta} (H(I)[\log(1/\delta)+Ent(I)]). $ We also give a single $\delta$-correct algorithm with expected sample complexity $ O ( H(I)[\log(1/\delta)+Ent(I)] +D\log(e+\log(e+D))),D=\Delta_{[2]}^{-2}, $ without prior knowledge of the gaps. Our lower bound removes the dyadic-gap and monotonicity restrictions of previous work, and our upper bound removes the additional polylogarithmic factor multiplying the two-arm term. Thus, a single algorithm attains the instance-wise lower bound up to an additive two-arm term. The main theorems have been formalized and proved in Lean 4.

---

347. Leakage-Safe and Scheduler-Aware Machine Learning for Grid Job Runtime Prediction

• Authors: Ashfaq Ali Shafin , Khandaker Mamun Ahmed
• URL: https://arxiv.org/abs/2609.13701
• Abstract:

  Accurate job runtime prediction can improve scheduling-aware resource management in grid and distributed computing environments, but prediction models must be evaluated under realistic deployment constraints. This paper revisits CPU burst time prediction on the GWA-T-4 AuverGrid workload trace and reformulates it as leakage-safe pre-execution job runtime prediction. We define the target as job-level runtime, use only submission-time attributes, exclude post-execution variables, and evaluate models under temporal and cold-start settings rather than relying only on random cross-validation. We compare standard regressors, chronological historical baselines, categorical encoding strategies, and CatBoost with native categorical handling. We further add temporal hyperparameter tuning, runtime predictability analysis, feature ablation, error analysis by job length, and a minimal scheduling simulation. After temporal-validation tuning, CatBoost achieves the strongest deployment-oriented result with R^2=0.239, MAE=27,019, RMSE=46,587, and LogMAE=2.646 on the held-out temporal test set. A single-server simulation over all 69,523 held-out temporal test jobs shows that prediction-informed SJF reduces average waiting time by 50.92% relative to FCFS. The results show that random-split evaluation overestimates performance, categorical-native boosting improves temporal generalization, and long-job underestimation remains a scheduler-relevant challenge.

---

348. Not all Negation Cues are Equal: Affixal Negations Yield Better Negation Understanding

• Authors: Tian Tan , Eduardo Blanco
• URL: https://arxiv.org/abs/2609.13685
• Abstract:

  Negation remains a longstanding challenge for both language models (LMs) and large language models (LLMs). Prior work mainly focuses on a small set of high-frequency single-word negation cues, such as not and never, with limited exploration of broader negation types and modern LLMs. To address this gap, we construct NegCue, a large-scale dataset containing over 1.8M samples spanning single-word, multi-word, and affixal negation with more than 200 unique cues. We further pre-train both encoder-only LMs and LLMs on NegCue to investigate how different negation types affect negation understanding. Experiments on five downstream benchmarks show that negation types contribute unevenly to performance gains under the same training scale. In particular, affixal negation yields the largest improvements, while the gains from the commonly studied single-word negation remain modest. Moreover, our results demonstrate that further pre-training improves negation understanding for both LMs and LLMs.

---

349. LayerRoute: Adaptive Layer-Skipping with LoRA-Preserved Quality for Efficient LLM Inference

• Authors: Prateek Kumar Sikdar
• URL: https://arxiv.org/abs/2609.13682
• Abstract:

  We introduce LayerRoute, a parameter-efficient method for adaptive transformer layer-skipping that combines per-layer hard-gated routing (trained via a straight-through estimator) with joint LoRA fine-tuning. LayerRoute augments each of the 24 transformer blocks in Qwen2.5-0.5B-Instruct with a lightweight per-layer router (~21.5K parameters) and LoRA adapters (rank 8, ~1.08M parameters), training both jointly under a gate-regularized language-modeling objective. Across 10 independently-seeded training runs, LayerRoute converges to an identical skip-pattern structure in every run - a consistent set of 9 middle layers (8-16) becomes skip-eligible in all 10 seeds - and delivers genuine, verified wallclock speedup in every run (1.02x-1.06x, mean 1.04x). Quality is preserved or improved in every configuration tested: joint LoRA adaptation yields a perplexity improvement over the unmodified backbone in all 10 seeds (mean delta = -1.16 and -1.11 across the two evaluation splits used). We further verify the router performs genuine, non-trivial per-input computation: gate decisions in skip-eligible layers change the actual skip/run outcome for 87-100% of held-out samples, confirming real input-dependent routing rather than a fixed pruning pattern. LayerRoute trains in under 7 minutes on a single A100 and adds negligible overhead beyond the routing decision itself. We report our full reproducibility methodology, including a systematic diagnostic investigation into what determines the router’s per-input decisions, as part of this work.

---

350. FaithfulBench: Does AI Counsel Uphold or Undermine the User’s Professed Faith?

• Authors: M Waleed Kadous , Benjamin Olsen , Walter Scheirer , Daniel D. Slate , Alexander Arnold , DZ Kalman
• URL: https://arxiv.org/abs/2609.13634
• Abstract:

  Do AI assistants help believers reason about moral dilemmas consistently with their faith? We present FaithfulBench, the first benchmark to score AI counsel across traditions by how well it adheres to the user’s professed faith. Scenarios are drawn from each tradition’s most respected texts, with the faithful answer known and applied by the judges as the standard. We test five frontier models under three conditions: the AI does not know the user’s tradition; it receives a one-line prompt identifying the user as a practicing adherent; or it receives a companion-counselor guide rooted in the tradition’s sources. Two judges score the initial response and whether the model caves or holds when pressured toward the answer the user wants. When the tradition is unstated, models counsel from a secular therapeutic default and every model fails some believers. Naming the faith wins a faithful first answer but not steadfastness; the guide improves both.

---

351. An Efficient and Modular Framework for Targeted Harm Mitigation in LLMS

• Authors: Roberto Campbell , Momin Abbass , Muneeza Azmat , Michal Ulewicz , Raya Horesh , Kristjan Greenewald , Rogério Abreu de Paula , Nathalie Baracaldo
• URL: https://arxiv.org/abs/2609.13624
• Abstract:

  Large Language Models (LLMs) are powerful zero-shot learners but remain prone to misalignment with human preferences, often producing biased, toxic, or otherwise harmful outputs. Existing alignment methods, while effective, are costly and tightly coupled to the model, limiting flexibility and scalability. We propose a modular correction framework that augments pretrained LLMs with Activated LoRA (aLoRA) adapters and a context-aware routing mechanism to eliminate harms from misaligned model responses. Our approach enables expert adapters to activate mid-sequence without invalidating the KV cache, allowing low-latency, targeted correction during generation. Each expert is trained to detect and mitigate specific harms, such as bias or toxicity. A learned router dynamically selects appropriate experts based on the models intermediate outputs. We demonstrate that our system improves alignment on standard safety benchmarks while preserving task performance, offering a lightweight and efficient path toward safer and more controllable LLM deployments.

---

352. Predictive audio representations for early detection and tracking of hidden dynamic objects

• Authors: Katerina Vinciguerra , Moritz Brandes , Danilo Hollosi , Letizia Marchegiani
• URL: https://arxiv.org/abs/2609.13595
• Abstract:

  Predicting potential dangers is core to safety. Forecasting the presence of other traffic agents is core to danger prediction. Occluded traffic agents challenge detection systems as they might become visible too late, leaving the autonomous vehicle too little time to identify, plan and act accordingly in a robust and safe way. Previous works proved that auditory perception, being omnidirectional and not constrained by a field-of-view, provides fundamental cues for early spotting of different road users, even when hidden by other vehicles or infrastructures. Yet, those contributions deal with scenarios with only one vehicle present, and they either identify the type of vehicle or estimate its direction of arrival. In this work, we move forward, and propose a multi-task system which, simultaneously, estimates the number of vehicles present, their type, and their direction of arrival. Our methodological contribution is a two-stage pipeline: a self-supervised pre-training stage inspired by the Joint- Embedding Predictive Architecture (JEPA) applied directly to multichannel raw waveforms, followed by supervised multi-task fine-tuning with a bidirectional LSTM and three classification heads. The pre-training stage trains the encoder without labels, pushing it to predict the latent representation of a future audio segment from its past context. To train and test our framework, since no suitable dataset was publicly available, we collected an ad-hoc one covering Non-Line-Of-Sight scenarios with multiple traffic agents simultaneously operating. Experimental evaluation shows that our method outperforms the state of the art; it also proves that our design choice allows the model to learn robust representations, which can be transferred to an unseen driving scenario, maintaining reasonable and stable performance.

---

353. mKernel: Fast Multi-GPU, Multi-Node Fused Kernels

• Authors: Ziming Mao , Yihan Zhang , Shawn Wei Chew , Shuang Ma , Costin Raiciu , Yang Zhou , Scott Shenker , Ion Stoica
• URL: https://arxiv.org/abs/2609.13585
• Abstract:

  Communication has become a bottleneck in distributed training and inference of large models. Overlapping communication with computation at the granularity of kernels, on separate streams, reduces only part of this communication cost. Fused kernels often have better performance by transmitting each output tile as soon as it is produced, but existing fused kernels are largely confined to a single NVLink domain. We present mKernel, a library of multi-GPU, multi-node fused kernels that overlap computation, intra-node NVLink communication, and inter-node RDMA at tile granularity. mKernel partitions the streaming multiprocessors (SMs) of a persistent kernel into compute and communication roles, and an on-GPU controller tunes the SM partition adaptively at run time, since the best SM partition varies with the kernel and the input shape. It structures data movement hierarchically so that data traversing the inter-node network is minimized. Finally, it drives the network from the GPU through a lightweight command queue and host proxy implemented directly on RDMA verbs, which allows the same kernels to run on any network backend (e.g. InfiniBand and on AWS EFA); we observe, surprisingly, that GPUDirect Async (IBGDA) yields little additional benefit over host-assisted GPU-initiated communication. We implement five kernels spanning tensor, sequence, and expert parallelism. On two 16-GPU H200 clusters, mKernel achieves speedups of up to 1.72x on GEMM+AllReduce and $1.88\times$ on Ring Attention.

---

354. Same Patient, Different Order: Action-Level Reliability of Clinical LLM Agents Under Repeated Runs

• Authors: Rohith Reddy Bellibatlu , Manpreet Singh , Zhoutian Han , Wenbin Zhang
• URL: https://arxiv.org/abs/2609.13582
• Abstract:

  A clinical agent benchmark can report the same verdict on identical inputs while the agent files a materially different order on each run. Such agents order tests, request medications and place referrals, yet benchmarks typically score one run per task and rarely ask whether identical inputs produce identical actions; MedAgentBench, the benchmark we use, scores a single attempt and says so. To measure this gap we introduce “same-input rerun”, which replays a task with every input held fixed and compares the orders rather than the score, with six reliability metrics, and apply it to 1000 MedAgentBench runs across 50 tasks from its five write-capable families, two open-weight models below ten billion parameters quantised to four bits, and two temperatures. The study establishes that action-level divergence exists and can pass unrecorded by the score, not that any rate generalises. Under the 8B model at temperature 0.7, all 43 ordering groups emit a different set of orders across five identical runs, 26 emit the order on some runs and not others, and 28 record a different coded value, dose or analyte. In 22 of those 43 the benchmark reports the same failing verdict for materially different behaviour, as it does for all 10 divergent groups of the 4B model at 0.7. Orders also reach different endpoints across runs, one of which the record server rejects while the agent is told it succeeded. These findings motivate repeated-run evaluation, action-level stability reporting and execution-faithful environment feedback in clinical-agent benchmarks.

---

355. Domain-Specific Jargon in Large Language Models: A Comparative Analysis between General-Purpose and Specialist Models

• Authors: Darin Keng , Zhewei Sun
• URL: https://arxiv.org/abs/2609.13556
• Abstract:

  Large Language Models (LLMs) have shown remarkable proficiency on general-purpose tasks, yet their performance often degrades in highly-specialized technical domains. Moreover, little is known about how parametric knowledge of domain-specific terms is encoded within these models. We address this gap by contributing two novel medical jargon evaluation benchmarks and evaluate a general-purpose Llama-3.1 model against a variant fine-tuned on medical-domain data. Surprisingly, the general-purpose model outperforms the medically fine-tuned model on both tasks. Using mechanistic interpretability tools, we find systematic patterns of miscalibration for the medically fine-tuned model. Instead of reorganizing parametric knowledge, the fine-tuned model places greater emphasis on a small subset of model components associated with jargon-favoring predictions. We find that applying component reweighting strategies against the benchmark tasks successfully suppresses these components and closes the gap with the general-purpose baseline. We also observe that some jargon-sensitive components transfer knowledge to the same tasks involving materials science jargon, suggesting they encode a partially domain-agnostic notion of specialized terminology. Our results provide a case study in which a medically fine-tuned checkpoint does not improve jargon comprehension over its general-purpose counterpart, highlighting that domain adaptation should not be assumed to yield better performance on specialized terminology.

---

356. Attention Is All You Need (to Avoid Spurious Oscillations)

• Authors: Jinyoung Jeong , Joseph B. Choi , Xinlun Cheng , H. S. Udaykumar , Sanghun Choi , Stephen S. Baek
• URL: https://arxiv.org/abs/2609.13531
• Abstract:

  Can attention move a shock across several cells in one update without breaking it? We develop a conservative, fixed grid finite-volume scheme in which a CFL-conditioned attention flux selects upstream information according to the transport required by the current time step. One-dimensional inviscid Burgers transport is used as the central mechanism test: the same learned flux remains reliable in the conventional small-step regime and, with a time step four times larger, preserves sharp shocks while using one stage per update. A standard fifth-order WENO scheme with third-order strong-stability-preserving Runge-Kutta time integration (WENO-5+SSP-RK3) is included alongside controlled Forward Euler comparisons to separate flux selection from time integration. The learned attention shifts upstream with the local transport reach and becomes more selective near shocks; inference-time interventions and retrained ablations show that transport-scale information and state-dependent selection contribute directly to performance. Directional two-dimensional scalar Burgers transport and the one-dimensional shallow-water system then test whether the conservation-scale-selection principle transfers beyond the original scalar setting. The results support attention as a learnable information stencil for conservative large-step shock transport, while identifying finite candidate reach and problem-dependent robustness as the present limits.

---

357. Generative Interpretability via Scalable Neuro-Symbolic Models

• Authors: Xiaocong Yang
• URL: https://arxiv.org/abs/2609.13529
• Abstract:

  As the use of Large Language Models moves from chatbots into agentic systems, where outputs become actions with irreversible consequences on reality, the existing paradigm on AI Interpretability research, post-hoc interpretability, is structurally inadequate for safe and trustworthy model deployment: it explains behavior after the fact but cannot audit or intervene in an inference computation before it commits to an output. We therefore argue for a shift toward \emph{generative interpretability}, an architectural property under which a model’s inference pass natively exposes semantically meaningful checkpoints that are human-understandable and amenable to causal intervention. We show the merits of generative interpretability as comparison to other interpretability research paradigms, and propose Neuro-Symbolic Models as a concrete instantiation.

---

358. Rolling Day-Wise Mortality Prediction in Critically Ill Patients With AKI on CRRT Utilizing Machine Pressure Waveforms

• Authors: Shehan Irteza Pranto , Joanna Yang , Joshua Lambert , Stuart L. Goldstein , Lili Chan , Girish N. Nadkarni , Tiago K. Colicchio , Javier A. Neyra , Jin Chen
• URL: https://arxiv.org/abs/2609.13524
• Abstract:

  Critically ill patients with acute kidney injury (AKI) on continuous renal replacement therapy (CRRT) face high mortality, yet current risk assessment relies primarily on clinical parameters from electronic health records (EHR) and ignores minute-level circuit pressure waveforms generated by CRRT machines that track the extracorporeal circuit’s interaction with the patient. Clinicians therefore cannot see deterioration as it develops. Risk is reassessed only when labs are drawn, while this continuous record is discarded because it is contaminated by shared-device records, non-physiological minutes, and sensor artifacts. To make the stream usable, we aligned machine records to charted therapy intervals to prevent cross-patient leakage, removed priming and downtime minutes, tuned denoising on a synthetic spike-injection benchmark, and masked unobserved intervals rather than imputing them. On this cleaned stream, we define a rolling day-wise task and a transformer-based stacked ensemble that late-fuses a window-reduced sequence transformer with classical models using circuit-instability features and clinical EHR variables. In a leak-safe benchmark on the multi-center CRRTnet cohort (976 patients, 4,585 treatment days), the machine-only model had the lowest standalone prognostic value (AUROC 0.625), followed by the EHR-only model (0.717). Integrating EHR and machine streams reached a one-day mortality AUROC of 0.766. SHAP attribution showed that circuit-instability descriptors raised the machine share of the top 15 combined-model features from 3 to 7 (20.0% to 46.7%), highlighting filter pressure, transmembrane pressure (TMP), and access-to-return difference (ARD). To our knowledge, this is the first patient-level mortality prediction incorporating CRRT machine data, turning a discarded bedside stream into a continuous risk signal.

---

359. Positive Topology and Feasible Refinement: Forcing Matrices, Positivity, and Information

• Authors: Mirco A. Mannucci , Giovanni Sambin
• URL: https://arxiv.org/abs/2609.13523
• Abstract:

  We develop a conceptual and operational account of Positive Topology starting from a basic relation between points or models and observable properties. From this relation, two complementary structures emerge. The first captures universal refinement and cover: what must hold across all relevant cases and how information can be systematically refined. The second captures positivity and witnessed existence: what can be positively realized and sustained without relying on classical complements. A central result shows that the underlying relation between points and observables can be reconstructed from either of these induced structures. We also clarify the distinction between point-based and pointfree formulations: when points are available, positivity can be derived from the underlying forcing relation, while in the formal pointfree setting positivity is taken as primitive and its compatibility with cover is imposed axiomatically. We develop two complementary interpretations of the framework. The first is information-theoretic, viewing cover as refinement of partial information and positivity as witnessed feasibility. The second is game-theoretic, viewing positivity as the ability of a witness or hypothesis to survive successive legitimate refinements. The final part of the paper is deliberately programmatic. We outline how resource constraints, verification costs, and finite budgets can be incorporated into the framework. This leads to resource-sensitive notions of forcing, cover, positivity, and refinement, and raises new questions about how these structures behave as available resources change. Examples from medical diagnosis, legal reasoning, and AI systems illustrate the potential relevance of the approach to grounded, explainable, and resource-aware inference.

---

360. Adaptive Phase-Switching for Communication-Efficient Federated LoRA Fine-Tuning

• Authors: Jerry Adams Franklin
• URL: https://arxiv.org/abs/2609.13512
• Abstract:

  Federated fine-tuning of large language models with low-rank adaptation reduces per-client trainable parameters, but client-to-server communication remains the dominant cost. Existing accounting for federated LoRA protocols omits the asymmetric transition round when a protocol changes aggregation mode, and reports savings that ignore grouped-query attention shapes. This paper measures per-round upload and download bytes for a bidirectional B-only federated LoRA protocol and places five methods, three from prior work, on a single communication-quality frontier. The frontier has a knee, which an adaptive phase-switching aggregator, ReverseAdaptive, locates by monitoring the relative improvement in global training loss against a dimensionless threshold rather than by fixing a phase boundary in advance. On TinyLlama-1.1B-Chat with Alpaca, ReverseAdaptive attains 40.5 percent measured round-trip savings over FLoRA at a held-out instruction-following loss cost of 0.0063. It outperforms FFA-LoRA, which freezes the first of the two LoRA factors at initialization, by 0.0182 in held-out loss, more than twenty times the largest per-method seed standard deviation on that metric, so learning that factor before freezing it produces better adapters. The same threshold transfers across model scales without retuning, and the quality cost of the transition is stable across the two datasets tested.

---

361. A Three-Axis Stress Test of LLM vs Classical ML for Network Intrusion Detection under Distribution Shift and Adversarial Evasion

• Authors: Muhammad Ebad Atif , Muhammad Haider Ali
• URL: https://arxiv.org/abs/2609.13511
• Abstract:

  Large language models are increasingly benchmarked against classical machine learning for network intrusion detection (NIDS), almost always using same-dataset evaluation, and that protocol turns out to be incomplete. Evaluating XGBoost and RoBERTa-LoRA on two independently collected NetFlow v2 networks across three axes (same-dataset performance, cross-dataset transfer, and adversarial evasion) reveals no universal winner. The two models are statistically tied same-dataset. XGBoost wins decisively under cross-dataset distribution shift, by 15 points of F1 and 25 points of balanced accuracy; on the target network RoBERTa-LoRA’s false positive rate reaches 0.78, leaving it barely above chance despite a superficially moderate F1. RoBERTa-LoRA wins decisively under adversarial evasion, by roughly 17 points of F1 at a representative mid-range perturbation strength, while both models hold false positive rates below 0.01 throughout. The model an evaluator would recommend therefore depends entirely on which axis is tested, not on same-dataset accuracy alone. A staged feature-leakage ablation improves cross-dataset transfer non-monotonically, indicating the leakage signal is distributed across the feature representation rather than confined to a few columns, and cross-dataset transfer between our two networks is strongly directional. These results argue for evaluating NIDS models along multiple independent robustness axes, and with more than one metric per axis.

---

362. One Spectrum, Two Resources: Data-Memory Scaling in Autoregressive Prediction

• Authors: Chiwun Yang , Xiaoyu Li
• URL: https://arxiv.org/abs/2609.13500
• Abstract:

  How much learned memory is needed to benefit from more data? We show that the two resources are governed by one predictive-energy spectrum in a positive-entropy autoregressive retrieval source. Each coordinate contributes its query probability times the squared radius of its unknown logit. Writing $\mu$ for the resulting energy spectrum, we prove the minimax law $\mathfrak R^*{\rm value}(n,B)\asymp_R \Phi\mu(n^{-1})+\Phi_\mu(\tau_B), \Phi_\mu(t)=\int\min{x,t}\,\mu(\mathrm dx),$ for $n$ prediction blocks and a learned state with at most $2^B$ values. Data set the resolution $1/n$; memory sets the level $\tau_B$ reached by optimal bit allocation. The complete curve also recovers the positive spectrum. Energy-dimension pairing is essential: two causal sources with identical block-energy and block-dimension marginals have different data and memory exponents. A masked query-key attention head learns the route and values, realizing the law with explicit routing, format, and arithmetic errors. Further results give exponent-adaptive allocation, finite-precision realization, and compute-precision laws under two-sided arithmetic assumptions. Experiments recover the data-memory collapse and coupling exponents, explain the routing and allocation mechanisms, and examine weight-only quantization across six pretrained-model scales.

---

363. Canaries in the Bank: Auditing User-Level Privacy in Private Evolution

• Authors: Sai Aparna Aketi , Enayat Ullah , Shripad Gade
• URL: https://arxiv.org/abs/2609.13499
• Abstract:

  Private Evolution (PE) generates high-fidelity synthetic data in federated settings without exposing users’ raw data. It aggregates clipped user votes over a shared candidate bank into a differentially private histogram, with noise calibrated to the worst-case user contribution. However, it is unclear whether an adversary can realize this worst-case privacy loss while following the PE protocol. We introduce a protocol-aware empirical audit in which the server commits to a single shared candidate bank and replaces roughly 1% of its entries with probes derived from a known, non-private canary. We evaluate eight attacks, including an unchanged-bank baseline, exact copies, plausible paraphrases, and high-entropy synthetic nonces. Experiments on Yelp and Sentiment140 show that natural-text attacks remain substantially below the theoretical DP bound, while nonce-based attacks yield considerably stronger bounds and come closest to the mechanism’s privacy ceiling. These results quantify the gap between formal worst-case privacy and leakage achievable through protocol-valid candidate-bank manipulation.

---

364. Building a Production Greek-English Speech Recognizer

• Authors: Christos Petrocheilos , Cleopatra Papadopoulou , Chris Porikis , Ioakeim Perros , Ayoub Kirouane , Themistoklis Nikolis
• URL: https://arxiv.org/abs/2609.13498
• Abstract:

  We report a multi-month engineering program to build Sophea, a production bilingual Greek-English automatic speech recognition system. We evaluate the system against nine production gates covering Greek and English word error rate, language identification, and hallucinations on non-speech audio. Across twenty-three training iterations and two model architectures, no training-data composition passed all nine gates simultaneously. Meeting the Greek noisy-environment target required about 1,500 steps of dense domain exposure, while preserving English language identification tolerated only about 250 steps, or about 1,250 with a rebalanced mix that reduced Greek accuracy. We describe a six-stage data pipeline in which calibrating an audio-quality filter against in-domain anchors reduced the discarded share of scored Greek audio from 98.7 percent to 10.6 percent. A pre-registered ablation isolated a hallucination defect to one training-data package. A three-model ROVER ensemble increased gate coverage from 4-7 of 9 for individual models to 9 of 9 and reduced overlapping-speech WER from 53.35 percent to 37.87 percent, a 29 percent relative improvement. A separate learned per-clip arbiter over two models is listed as sophea/asr-k1 (preview) on the public Open ASR Leaderboard, with 4.26 percent average WER across eight public English test sets, and reaches 25.88 percent WER on live Greek noisy-environment traffic. We also document five cases in which a measurement tool produced a plausible but incorrect result and seven substantial approaches that were evaluated but not shipped. No model weights or training data are released; we report methodology and quantitative results only.

---

365. Generative AI and Extended Reality in Collaborative Architectural Design Education: An Exploratory Studio Study

• Authors: Yao Xiao , Max Chen , Yichen Li , Nathaniel Powers , Maxwell Wiesenfeld , Gillian Smith , Soroush Farzin , Shichao Liu
• URL: https://arxiv.org/abs/2609.13494
• Abstract:

  Architectural design education relies heavily on visual ideation and representation to support collaborative learning in studio environments. Recent advances in generative artificial intelligence (GenAI) and extended reality (XR) offer new opportunities for rapid idea exploration and immersive spatial visualization. This exploratory mixed-methods classroom study investigated how GenAI-assisted multi-user XR influenced collaborative architectural conceptual design. We developed GenARch, a pipeline that integrates GenAI-based visual generation with collaborative XR environments, and deployed it in an undergraduate architectural design studio. Twenty-seven students formed seven self-selected design teams; four teams incorporated GenARch into their usual course workflow to support collaborative ideation and visualization, while three teams continued the same course workflow without GenARch. Pre- and post-intervention surveys assessed design self-efficacy, attitudes toward collaborative learning, and teamwork; a seven-member panel evaluated team design presentations; and GenARch teams participated in group interviews. The quantitative results showed larger relative declines in confidence and outcome expectancy for the GenARch condition and a positive difference-in-differences estimate for perceived conflict management, while panel-rated presentation outcomes were not significantly different between conditions. Interviews indicated complementary roles for the technologies: GenAI supported idea externalization and visual reference generation, whereas XR supported spatial, contextual, and scale-based evaluation. Students also reported challenges related to control, dimensional fidelity, shared attention, and motion comfort. These findings highlight both opportunities and limitations when GenAI and XR are incorporated into collaborative design education.

---

366. Edge-addition monotonicity of positive p-energy fails for every p >= 1

• Authors: Koyar Afrasyab
• URL: https://arxiv.org/abs/2609.13476
• Abstract:

  At a 2021 AIM workshop, Guo conjectured that the positive square energy s+ = E+_2 should inherit the familiar edge-addition monotonicity of the spectral radius, rho(G + uv) >= rho(G). That conjecture was subsequently shown to fail at p = 2. Tang, Liu, and Wang then introduced positive p-energy, proved nonmonotonicity for every 1 <= p < 3, and in version 3 of their preprint (26 March 2025) explicitly conjectured that monotonicity should hold for p >= 3. We disprove this conjectured high-exponent extension completely: for every real p > 2 there are infinitely many connected graphs G and nonedges uv such that E+_p(G + uv) < E+_p(G). Together with the Tang-Liu-Wang counterexamples below 3, this shows that no exponent p >= 1 restores the spectral-radius-style monotonicity: positive p-energy can decrease under the addition of an edge for every real p >= 1. The construction is a chain of clique blocks joined by regular bipartite graphs. An equitable quotient converges to Q = I + cA(P_k). For noninteger p, a binomial-series sign argument for a fractional power of I - cA(P_k) gives the required negative endpoint entry. At integer exponents, choosing c across the first spectral threshold leaves exactly one negative eigenvalue, and path locality forces the positive spectral contribution to have negative sign. We also give a fully rational 38-vertex certificate at p = 4 and determine the complete failure interval of a fixed 17-vertex counterexample at p = 3.

---

367. Symmetry- and Property-Aware Crystal Generation with Reinforcement Learning for Inverse Materials Design

• Authors: Ting-Wei Hsu , Arun Bansil , Qimin Yan
• URL: https://arxiv.org/abs/2609.13468
• Abstract:

  The inverse design of crystalline materials ultimately seeks structures with desired physical properties. However, for many functional responses, a favorable numerical value is meaningful only when supported by the symmetry of the underlying crystal. Without the appropriate crystallographic constraints, an apparent response may be ill defined, accidental, or not symmetry protected. Here we introduce SPARC, a symmetry- and property-aware reinforcement learning framework that optimizes physical objectives while preserving the structural conditions required for their realization. We demonstrate SPARC on two complementary tasks. The first targets strong uniaxial dielectric anisotropy, a tensorial response that is well defined only within appropriate crystal classes. The second maximizes the spectroscopic limited maximum efficiency, a scalar device-level objective without a prescribed symmetry class, allowing the framework to identify favorable crystallographic motifs. These results show that symmetry is not merely an additional design constraint, but a physical foundation for generating candidates with meaningful, robust, and realizable functional properties.

---

368. Hindsight Bias in Clinical Temporal Reasoning: How Future Data Exposure Affects Large Language Model Judgment

• Authors: Misaki Matsuura , Sayantan Kumar , Ojas Kadam , Jeremy C. Weiss
• URL: https://arxiv.org/abs/2609.13454
• Abstract:

  Clinical decisions are prospective, but clinical language models are often evaluated on retrospective records that reveal the final diagnosis, treatment response, and outcome. Such evaluations may reward the use of future information rather than reasoning under the uncertainty present at the decision point. We introduce a paired benchmark for measuring outcome-conditioned shifts consistent with hindsight bias in clinical temporal reasoning. It contains 171 case reports from the PubMed Central Open Access Subset—40 sepsis and 131 GLP-1/diabetes cases—represented as both textual narratives and human-annotated and LLM-generated textual time series (TTS). For each case, questions are tied to a clinically meaningful cutoff and paired with a prospective reference answer and an outcome-consistent \emph{hindsight trap}. Models answer each question using either a TTS truncated at the cutoff or the complete timeline; additional conditions vary the narrative source (original or synthetic) and TTS annotation source (human or LLM). We evaluate accuracy (Acc), hindsight trap rate (HTR), answer instability rate (AIR), and hindsight bias rate (HBR), each of which captures different signals of hindsight bias. Across GPT 5.6 Sol, Gemma 4, GLM 5.2, and Opus 5, full timeline exposure produces consistent hindsight-sensitive shifts, while temporal masking reduces bias without lowering accuracy.

---

369. Learning to Solve Hard Problems in RL for LLMs by Never Giving Up

• Authors: Michael Noukhovitch , Hamish Ivison , Nathan Lambert , Aaron Courville
• URL: https://arxiv.org/abs/2609.13443
• Abstract:

  We demonstrate that training LLMs with RL does not improve performance equally across a dataset. RL shows large improvements on easy problems that an LLM is already good at solving, but small improvements on hard problems. We call this the Matthew Effect in RL for LLMs, after the phenomenon of cumulative advantage from economics and network science summarized as “the rich get richer”. The naive explanation is that hard problems require more compute to find a solution. We argue that modern RL methods are exacerbating the issue by wasting too much compute on easy problems and instead should dynamically reallocate how they use compute. We introduce Never Give Up (NGU), a simple adaptive sampling method that keeps generating samples for a problem until one is correct. By leveraging asynchronous RL, this naturally uses fewer samples to filter out easy problems and allocates more compute to solving harder problems. We investigate the design choices that affect NGU, such as off-policy robustness, and develop a set of best practices. On the math benchmark Deepscaler, NGU improves performance per compute, especially on harder problems. On a recent coding task, Manufactoria, standard GRPO with a per-test reward fails to fully solve problems that have a range of easy and difficult tests. NGU iteratively improves, solving harder and harder tests, until it learns to fully solve coding problems.

---

370. Certifiably Interpretable Training of ReLU-MLPs for Boolean Tasks with Guaranteed Truth-Table Generalization

• Authors: Hrad Ghoukasian , Anastasis Kratsios
• URL: https://arxiv.org/abs/2609.13439
• Abstract:

  As compute scales, models evolve, and training algorithms advance, our ability to explain the increasingly powerful AI systems they enable is eroding. To help safeguard interpretability, we introduce a specialized training algorithm (MACCHIATO) that jointly constructs (i) an explicitly structured $\operatorname{ReLU}$-MLP from partial truth-table observations and (ii) an explicit Boolean circuit over signed literals with ${\operatorname{AND},\operatorname{OR},\operatorname{XOR}}$ gates certifying what its subnetworks compute and how they compose. Intuitively, we iteratively project the residuals of a Boolean function onto low-dimensional ${\operatorname{AND},\operatorname{OR},\operatorname{XOR}}$-circuit classes and exactly compile the resulting circuit into a $\operatorname{ReLU}$-MLP; we combine $\operatorname{ReLU}$-MLP circuit compilation, ESPRESSO logic minimization, and influence-based variable selection. Roughly speaking, our interpretability certificate is complemented by a statistical guarantee: under the theorem’s influence-recovery conditions, if each of the $m$ stage-wise residuals depends on at most $\log_2(B)$ bits, a sample-splitting variant of our algorithm trained on $T$ observations returns a six-layer $\operatorname{ReLU}$-MLP (counting the input layer) of width $\mathcal{O}(mB)$ with truth-table error $\mathcal{O}\bigl(\sqrt{m(B+\log(m/\delta))/T}\bigr)$. On synthetic random-junta tasks, our networks outperform depth- and hidden-width-matched Adam-trained MLPs in several data-sparse or projection-aligned regimes, while the trained ReLU-MLPs are stronger in others. Moreover, in our explicit PyEDA truth-table implementation, the iterative procedure completes in regimes where flat ambient-dimensional ESPRESSO exceeds the three-hour computational budget.

---

371. Chance-Constrained Belief-Space Maneuver Planning for Autonomous Collision Avoidance Under Uncertainty

• Authors: Grace Ra Kim , Duncan Eddy , Mykel J. Kochenderfer
• URL: https://arxiv.org/abs/2609.13428
• Abstract:

  Increasing conjunction frequency in low Earth orbit places growing pressure on spacecraft operators to determine not only whether an encounter requires mitigation, but whether sufficient information is available to commit to a maneuver. This work formulates this information-action tradeoff as a belief-space planning problem for conjunctions between a maneuverable spacecraft and an unmaneuverable secondary object. The planner represents the uncertain orbital states as Gaussian beliefs and uses a chance-constrained belief-space Monte Carlo tree search framework to reason over possible future tracking updates before time of closest approach (TCA). A terminal chance constraint limits the probability of reaching TCA above a prescribed collision-risk threshold, allowing the planner to wait for informative tracking while intervening when deferral becomes too risky. We evaluate the approach on eight historical conjunctions from NASA’s Conjunction Assessment Risk Analysis dataset. By varying the secondary-object measurement quality and tracking cadence, we generate a total of 96 distinct evaluation scenarios. Across the evaluated conditions, the planner reaches TCA without maneuvering in approximately 40% of episodes while maintaining no terminal collision-risk violations. In contrast, fixed-time rule-based maneuver policies resolve more encounters without maneuvering when intervention is deferred closer to TCA, but at the expense of increasing terminal risk violations. The fraction of episodes reaching TCA without maneuvering depends strongly on tracking quality and measurement cadence, ranging from 76% under accurate, frequent measurements to approximately 18%-20% under the poorest tracking conditions. These results show that tracking quality and frequency are not only inputs to collision-risk estimation: they can determine when intervention becomes necessary.

---

372. ReCAST: Reward Credit Assignment across Timesteps for Online Diffusion Reinforcement

• Authors: Yihang Chen , Yuanhao Ban , Kuei-Chun Kao , Cho-Jui Hsieh
• URL: https://arxiv.org/abs/2609.13425
• Abstract:

  Training diffusion models with multiple rewards requires distinguishing user preference from reward informativeness. User preference determines how much each reward should contribute to the overall objective; reward informativeness determines when its feedback is useful during denoising. Some rewards can meaningfully evaluate a sample as soon as global structure emerges, but others become informative only when the sample is nearly clean. To address both questions jointly, we propose ReCAST (Reward Credit ASsignment across T}imesteps), the first method, to our knowledge, for per-reward, timestep-dependent credit assignment in diffusion reward fine-tuning. ReCAST separates user preferences from temporal allocation through a reward-by-timestep weight matrix $W$, whose row sums match the user-specified reward budgets $\lambda$, while its column sums are equal, assigning the same total weight to each denoising step. Under these marginal constraints, ReCAST allocates weight according to each reward’s informativeness, quantified by its Rényi discriminability gain at each step. These gains telescope to the total discriminability between the reward-induced positive policy and the current policy, providing a basis for temporal credit assignment. We evaluate ReCAST by training SD3.5-Medium under two distinct four-reward settings, each across five reward budgets $\lambda$. ReCAST improves the training rewards in one setting and matches them in the other, improves every held-out judge in both, and is preferred by an independent LLM-as-a-Judge. Together, these results show that ReCAST yields improvements that generalize beyond the training rewards and support its core principle: assigning each reward greater weight at the denoising timesteps where its feedback is most informative.

---

373. Task-Aware Federated Fine-Tuning for MoE-based Large Language Models

• Authors: Tingqi Wang , Hongyu Ke , Haoxin Wang , Rafal Angryk , Zhipeng Cai
• URL: https://arxiv.org/abs/2609.13395
• Abstract:

  Mixture-of-Experts (MoE) has become a widely adopted architecture for Large Language Models (LLMs), as it improves model capacity while limiting computational overhead through sparse expert activation. This property makes MoE-based LLMs particularly attractive for resource-constrained distributed environments. However, federated fine-tuning of MoE-based LLMs remains challenging under heterogeneous client data. Since clients often correspond to different task preferences, directly aggregating their local updates may weaken expert specialization and introduce conflicting update directions on shared experts. To address these challenges, we propose FedTAR, a task-aware federated fine-tuning method for MoE-based LLMs. FedTAR establishes the association between local updates and task preference via routing outputs. Specifically, we apply Singular Value Decomposition (SVD) to both routing features and local updates to extract low-dimensional task coordinates and update directions. Based on the task coordinates, FedTAR performs intra-cluster aggregation among clients with similar task preferences and inter-cluster aggregation across different task groups. The aggregated update is then reconstructed through the learned task-to-update mapping, ensuring that the final update remains aligned with task-specific optimization directions. In this way, FedTAR preserves expert specialization and mitigates destructive interference among heterogeneous clients. We evaluate FedTAR on four benchmark tasks under different non-IID settings. Experimental results demonstrate that FedTAR consistently outperforms strong federated fine-tuning baselines and achieves state-of-the-art performance.

---

374. SkillAtlas: An Attack Trace Library for Agent Skills

• Authors: Yuxin Tian , Zenghao Duan , Liang Pang , Zhiyi Yin , Xueqi Cheng
• URL: https://arxiv.org/abs/2609.13353
• Abstract:

  Agent skills are reusable units for language-model agents, but their risks emerge through model decisions, user context, tool calls, and execution feedback rather than through stable signatures or a single sandbox run. Existing static, dynamic, and benchmark-style evaluations rarely preserve public evidence that can be inspected, searched, and reused. We present SkillAtlas, a hosted attack trace library that converts private agent-skill security report bundles into reviewed, redacted, and searchable public cases. The library contains 3,014 cases, 6,589 traces, 151,131 steps, 233 affected skills, and 8 risk categories; 42.5% of successful cases first become successful after a non-success initial round, and trajectory-grounded labels improve pre-execution guard accuracy to 0.770.

---

375. ProtoCAM: Interpretable Few-Shot Mask-Guided Prototypical Learning for Breast Lesion Classification in Ultrasound Imaging

• Authors: Ashkan Ebadi
• URL: https://arxiv.org/abs/2609.13340
• Abstract:

  Breast ultrasound imaging plays an important role in the early detection and diagnosis of breast cancer, particularly for patients with dense breast tissue. However, developing reliable deep learning models for ultrasound analysis is challenging due to limited annotated medical data and the need for interpretable predictions. To address these challenges, this paper proposes ProtoCAM, an explainable few-shot learning framework for breast lesion classification that integrates mask-guided feature encoding, prototypical metric learning, and gradient-based visual explanations. The proposed approach leverages lesion masks to guide feature extraction and constructs class prototypes within an embedding space to enable robust classification under limited training samples. The framework was evaluated on the BUSI dataset using a stratified group k-fold cross-validation protocol to prevent patient-level data leakage. Experimental results demonstrate ProtoCAM’s high performance in low-data scenarios. In a 3-way 5-shot setting, the proposed method achieves a macro F1-score of 0.910, representing a substantial improvement over standard supervised CNN models. Among the evaluated backbone networks, ResNet18 achieved the best performance, reaching a macro F1-score of 91.65% under a 15-shot configuration, providing interpretable insights into the classification decisions. These results highlight the potential of explainable few-shot learning frameworks for reliable computer-aided breast cancer diagnosis in data-scarce medical imaging environments.

---

376. Bridging Thought and Action: Taming Long-Horizon Instability in Open-Source LLM Agents with a MetaTool-Enhanced ROS Framework

• Authors: Kazi Abrar Mahmud , Nilotpaul Kundu Dhurubo , Tamal Kirttonia , Sabbir Hossain Ujjal , Mohammad Ariful Haque
• URL: https://arxiv.org/abs/2609.13335
• Abstract:

  Large Language Models (LLMs) have enabled more natural human-robot interaction, but open-source models often exhibit unstable long-horizon reasoning and inefficient action execution when deployed in agentic robotic frameworks. This paper presents an enhanced ROS-Agent based architecture that improves task reliability and execution efficiency for agentic robotic systems using open-source LLMs. The proposed system introduces a novel intermediate mechanism, termed the MetaTool, which enforces structured planning prior to action execution. Given a natural-language command, the MetaTool induces the LLM to generate a pseudo-code plan of intended tool invocations, which is stored in the ROS-Agent’s scratchpad and persists throughout execution. By explicitly separating planning from execution, the proposed approach reduces execution loops and improves deterministic behavior. The architecture is validated on a custom mobile robotic platform with multimodal perception and motion control capabilities. Experimental results on real-world interactive tasks demonstrate improved task completion and contextual consistency, with up to ~24% gains on complex tasks compared to the baseline framework.

---

377. The Agentic Company OS: Substrate Inversion for Sustained Enterprise Agent Deployment

• Authors: Oliver Aleksander Larsen , Mahyar T. Moghaddam
• URL: https://arxiv.org/abs/2609.13334
• Abstract:

  Enterprise AI agents often succeed in a demonstration and then stall once they must operate day after day. An industry report estimates that most pilots never reach production and that deployed systems rarely retain feedback or improve over time, while agent benchmarks show single-run successes masking unreliable repetition. We argue that these failure modes share a common architectural root: agents reason over data structured for human operators and traditional applications, not for the language models that power them. This position paper proposes that companies deploying agents in sustained operation should rebuild their cognitive substrate, the shared environment agents read as working context, around representations matched to that reasoning surface, isolating schema translation to the action boundary. Markdown is the instantiation available today, not a proven agent-native primitive. Two mechanisms ground the argument: context-bandwidth asymmetry, the gap between one-pass reading of connected prose and field-by-field typed access that strips relations; and cross-loop coupling, the claim that action, skill, and policy loops compound only if they share one substrate. A four-layer framework (Data, Knowledge, Intelligence, Governance) operationalizes the position, with a Sync Agent enforcing the action boundary and a per-skill trust gradient, making governance and auditability structural properties of the substrate. The position revives the shared-substrate tradition of classical multi-agent systems under LLM-era economics. We analyze the main objections and risks, including indirect prompt injection on the compile path, and outline a research agenda for evaluating substrates directly.

---

378. Pedestrian Crossing Intent Classification From Event-Based Vision Using Convolutional Spiking Neural Networks With Temporal Augmentation

• Authors: Henok Teklu , Mustafa Sakhai , Maciej Wielgosz , Matej Mertik
• URL: https://arxiv.org/abs/2609.13328
• Abstract:

  Anticipating whether a pedestrian will cross the road is safety-critical for autonomous vehicles, requiring real-time inference under challenging conditions including motion blur, high dynamic range, and class imbalance. Conventional frame-based deep networks process redundant RGB data at fixed frame rates, limiting their temporal resolution and energy efficiency. In this work we present an end-to-end pipeline that (i) converts real-world driving footage from the Joint Attention in Autonomous Driving (JAAD) dataset into synthetic dynamic vision sensor (DVS) event streams using the v2e simulator, (ii) augments training with the CARLA-simulated DVS sequences of the DVS-PedX dataset under both normal and adverse weather conditions, and (iii) trains a novel convolutional spiking neural network (Conv-SNN) with clip-consistent DVS augmentation to classify pedestrian crossing intent as binary: crossing or non-crossing. We detail all architectural decisions, the exact leaky-integrate-and-fire neuron dynamics with surrogate-gradient learning, the class-balanced loss formulation, JAAD oversampling at 6x, and a 70/15/15 stratified splitting protocol. The trained model achieves 95.83% accuracy and F1 = 0.9695 on the JAAD DVS test set, 97.79% accuracy and F1 = 0.9478 on normal CARLA DVS, and 94.78% accuracy and F1 = 0.8369 on adverse-weather CARLA DVS, all from a 1.07M-parameter architecture trained on CPU. Compared to prior frame-based approaches on JAAD, our method closes or surpasses the reported accuracy while operating natively on sparse temporal representations. We include a thorough analysis of the convergence behaviour across all 15 training epochs, domain transfer characteristics, and a quantitative comparison with representative related work.

---

379. Decoupling Error Attribution in Cloud-Native Graph-RAG: A Data Integrity Diagnostic Framework

• Authors: Shuai Yan , Yuhang Wu , Xiaodong Huang , Ke Wang
• URL: https://arxiv.org/abs/2609.13324
• Abstract:

  Graph-RAG systems often assume pristine data quality, overlooking the severe impact of perturbations in cloud-native databases. This paper proposes a three-layer decoupled diagnostic framework to orthogonally attribute system errors to reasoning loss, Knowledge Graph (KG) defects, and Cypher generation errors. Evaluated on a spatio-temporal ecological KG of the Southeastern Tibet region with eight defect types, results reveal that data integrity, rather than algorithmic reasoning, is the dominant performance bottleneck, with structural defects degrading system accuracy from 0.93 to 0.39. Crucially, we observe a masking-like phenomenon termed the Parametric Knowledge Masking Effect (PKME), suggesting LLMs compensate for broken retrieval paths using internal memory. This shrinks apparent query generation errors by over 70 percent, obscuring actual storage deterioration and increasing the risk of false negatives for automated monitoring. This work provides a quantitative foundation for auditing and optimizing data integrity in cloud-based information fusion systems.

---

380. Feasibility and Memory Mechanisms of Chern-Simons Context Reservoir Computation

• Authors: Jyotiranjan Beuria , Venkatesh H. Chembrolu
• URL: https://arxiv.org/abs/2609.13315
• Abstract:

  We investigate whether a Chern-Simons (CS) context reservoir is a viable computational substrate and whether evolving its gauge connection provides a benefit beyond simpler mechanisms. The reservoir state is a density fluctuation on a two-dimensional context manifold, whose drift is generated by a density-sourced connection. To separate generic reservoir behavior from gauge-specific effects, we compare four matched models: reciprocal transport, instantaneous transverse reconstruction, local nonlinear feedback, and fully coupled conserved-current CS dynamics. Across ten random seeds, the fully coupled CS dynamics propagates Gauss law to numerical precision, converges under spatial and temporal refinement, remains stable under constraint-compatible noise, and satisfies the spatial CS equation more accurately than the instantaneous controls. All four models exhibit fading scalar memory and distinguish matched pulse-order histories in density, with no resolved general advantage for coupled CS. The distinction appears in the flow geometry: coupled evolution supports circulating and longitudinal history channels simultaneously, retains them briefly after input removal, and yields a combined-feature pulse-order accuracy of (0.879\pm0.035), compared with (0.679\pm0.065) for the instantaneous-transverse control. The evolved connection also cannot be reconstructed from an instantaneous density snapshot or replaced by a fitted local multiplier. We therefore find a task-specific advantage for geometry- and order-sensitive processing, rather than generic reservoir superiority. Here ``topological’’ refers to the gauge organization of the state; the reported memory and cyclic-lag measures are not topological invariants.

---

381. Task-Based CT Protocol Optimization Using Reinforcement Learning and Virtual Imaging Trials

• Authors: Jiaqi Zou , David Fenwick , Vahid Tarokh , Nicholas Felice , Jayasai Rajagopal , Anuj Kapadia , Ehsan Samei , Navid NaderiAlizadeh , Ehsan Abadi
• URL: https://arxiv.org/abs/2609.13309
• Abstract:

  Protocol optimization in computed tomography (CT) aims to improve diagnostic image quality while reducing radiation dose, but the interdependence of acquisition and reconstruction parameters makes exhaustive testing impractical. We propose a virtual imaging trial framework with reinforcement learning for efficient CT protocol optimization. Sixty-three computational human models with liver lesions were imaged using a validated CT simulator across 468 combinations of acquisition and reconstruction parameters, including tube voltage, tube current, reconstruction kernel, slice thickness, and pixel size. The optimization objective balanced liver lesion detectability, quantified by detectability index d-prime, against radiation dose. A Proximal Policy Optimization agent was trained and conditioned on patient-specific CT localizer embeddings derived from a pretrained vision transformer. On held-out patients, evaluating only 8 protocols per patient, about 2% of exhaustive testing, recovered 98.2% of the exhaustive-search oracle objective. With no patient-specific simulation, surrogate scoring alone achieved 89.7% recovery. Conditioning on the localizer improved zero-simulation recovery by 10.7 percentage points over the localizer-blind policy (paired 95% CI 2.9-19.5; p=0.02). These results show that the proposed framework can substantially reduce exhaustive protocol testing while enabling task-based, dose-aware protocol selection before the diagnostic scan.

---

382. IMM-based Multiple Object Tracking using a State Prediction Neural Network

• Authors: Chan-Bin Lim , Dong-Hee Paek , Seung-Hyun Kong
• URL: https://arxiv.org/abs/2609.13307
• Abstract:

  Object tracking is essential for autonomous vehicles to avoid obstacles and plan routes. Radar maintains detection performance even in adverse weather and can measure relative velocity through the Doppler effect, making it well suited for object tracking. In this paper, we propose a data-driven state PRedictor-based Interacting Multiple Model tracking method (PR-IMM) that improves nonlinear object-motion representation while preserving the stability and interpretability of physics-based motion models. The proposed method employs a transformer-based PRediction model (PR) that incorporates radar Doppler measurements to predict object displacement. The PR model is integrated into the IMM as a mode alongside the CV, CA, and CT motion models, and their prior positions are dynamically combined according to the mode probabilities. Experimental results show that PR-IMM reduces position-estimation error by 57.3% over the IMM and by 16.5% over the PR, while reducing ID switches by 25.3% and improving IDF1 by 9.6% over the IMM.

---

383. Adaptive Conformal Redistribution for Inter-class Transitional Uncertainty in Medical Image Classification

• Authors: Saibal Ghosh , Samarup Bhattacharya , Sanjoy Kumar Saha , Umapada Pal , Tapabrata Chakraborti
• URL: https://arxiv.org/abs/2609.13303
• Abstract:

  Medical image classification is frequently complicated by transitional categories whose feature distributions overlap those of adjacent classes, producing ambiguous decision boundaries. Conformal prediction returns uncertainty-aware prediction sets, but these are not directly actionable in clinical screening, where a single decision is required. This work proposes adaptive conformal redistribution (AdaConRed), a label-free post-conformal decision rule that converts ambiguous prediction sets into refined class assignments. A five-stage pipeline is developed. Vision-language generative augmentation addresses minority-class scarcity; a frozen DermFoundation encoder provides embeddings; a lightweight multi-layer perceptron performs classification; an entropy-modulated, margin-aware nonconformity score constructs adaptive prediction sets; samples predicted as transitional with multi-label sets are reassigned to the most probable alternative class within the set, using only model outputs at inference. Evaluation uses the OSCC oral lesion and ISIC skin lesion benchmarks at a miscoverage level of 0.2. On the 3-class OSCC benchmark, overall accuracy improves from 73.54% to 77.38%, with oral cancer accuracy rising from 64.29% to 82.14% and benign accuracy from 56.57% to 70.20%. Reassignment of transitional samples reduces OPMD accuracy from 84.78% to 80.16%, consistent with the asymmetric cost of missed malignancy. On ISIC, overall accuracy improves from 85.83% to 87.19%, melanoma accuracy rising from 66.04% to 68.34%. AdaConRed outperforms LAC, APS and RAPS under an identical backbone and redistribution rule. Conformal prediction can be extended beyond uncertainty quantification toward actionable decision support where transitional disease categories are present, with gains concentrated in the clinically critical malignant categories. Code repository: this https URL .

---

384. Variational Template Matching with Statistical Fusion for Anomaly Detection in Patterned Structures

• Authors: Qinwu Xu , Yifan Jiang
• URL: https://arxiv.org/abs/2609.13298
• Abstract:

  Anomaly detection in structured images is challenging in small-data settings where deep learning approaches are costly or impractical. Classical template matching is simple and interpretable but lacks robustness to geometric variations such as scale, rotation, and perspective. We propose a variational template matching framework that represents anomaly templates as a family of transformed instances and performs detection via normalized cross-correlation over this transformation space. To further improve robustness, we introduce a density-based statistical anomaly score derived from local intensity distributions using kernel density estimation (KDE). This produces a smooth representation that captures distributional concentration and tail behavior more robustly than histogram-based methods. The structural and statistical signals are integrated through a unified fusion formulation, enabling complementary modeling of geometric similarity and distributional deviation. Experiments on biological cell images demonstrate that the proposed method outperforms classical baselines and achieves competitive performance with ResNet-50 under a fully training-free setting, while providing explicit localization. The approach offers an efficient, interpretable, and practical solution for anomaly detection in structured image domains.

---

385. LLaDA-UI: Bringing Block-wise Diffusion to Vision-Language GUI Agents

• Authors: Zhangxuan Gu , Haoxing Chen , Qi Qin , Yi Xin , Kai Gan , Lin Liu , Long Cui , Xiaomei Wang , Beitong Zhou , Yunzhu Zhang , Zhengwen Zeng , Changlong Gao , Weizhi Chen , Rongchao Zhang , Haoyuan Wu , Shuheng Shen , Changhua Meng , Weiqiang Wang , Jianguo Li , Zhenzhong Lan
• URL: https://arxiv.org/abs/2609.13287
• Abstract:

  Diffusion large language models (dLLMs) achieve high decoding efficiency through block-parallel, arbitrary-order generation, making them attractive for latency-sensitive applications. GUI agents represent a natural testbed for this paradigm, as they must repeatedly perceive screen states and emit structured, spatially grounded actions in real time. However, whether dLLMs can be extended into capable multimodal GUI agents while preserving their parallel decoding advantage remains an open question. We present LLaDA-UI, a 16.7B-parameter MoE-based, block-wise diffusion vision-language GUI agent. LLaDA-UI follows a two-stage training pipeline: general multimodal pre-training aligns a native-resolution vision encoder with the LLaDA2.0-mini-base diffusion language backbone, followed by GUI-agent supervised fine-tuning on diverse mobile, desktop, web, and grounding data. Across widely adopted grounding benchmarks and navigation benchmarks spanning multiple platforms, LLaDA-UI substantially outperforms Qwen2.5-VL-7B and surpasses Qwen3-VL-8B on four of six reported GUI benchmarks. These results establish block-wise diffusion as a practical generative paradigm for multimodal GUI agents.

---

386. Multimodal-Multiresolution Foundation Model for Lunar Remote Sensing

• Authors: Paolo Fraccaro , Gabby Nyirjesy , Daniela Szwarcman , Himanshu Patil , Vishal Gaur , Rohit Lal , Rachel A. Slank , Geoffrey Dawson , Hiyam Debary , Michael K. Barker , Andrew Annex , Vishnu Viswanathan , Zachary Morse , Ethan I. Schaefer , Nikolaos Dionelis , Ankur Kumar , Campbell D. Watson , Manil Maskey , Rebekah I. Dawson-Rigas , Juan Bernabé-Moreno , Rahul Ramachandran , Sujit Roy
• URL: https://arxiv.org/abs/2609.13283
• Abstract:

  We present a multimodal foundation model for lunar remote sensing, pretrained from scratch on SomBench, a geographically partitioned corpus of nearly two million co-registered tile bundles spanning 11 modalities at two spatial scales (1 m/pixel and 100 m/pixel). The model adapts the TerraMind masked-token architecture with two lunar-specific extensions: acquisition geometry is provided as explicit context, and meter- and hundred-meter-scale tiles are trained jointly so that a single set of weights covers both resolutions. FlexiViT patch embeddings allow adaptation to different patch sizes without retraining, while modality-wise inputs enable flexible multimodal fine-tuning. Qualitative generation experiments suggest the model learns meaningful cross-modal correspondences, including terrain derivatives from elevation and illumination-consistent reflectance from geometry. We evaluate on four benchmarks: crater detection at WAC and NAC scales, irregular mare patch (IMP) segmentation, and polar ice prospectivity regression. Across tasks, the pretrained model matches or outperforms ImageNet-pretrained baselines and an architecturally identical random-init control. On multimodal ice prospectivity regression, pretrained variants achieve the best results, while the random-init model outperforms most baselines, suggesting gains arise from both the architecture and pretraining. Label efficiency is notable for WAC crater detection, where the pretrained model trained on 50% of the data exceeds the strongest ImageNet baseline trained on the full dataset. Among adaptation strategies, LoRA matches or surpasses full fine-tuning on crater detection and IMP segmentation while using far fewer trainable parameters, whereas full fine-tuning performs best for ice prospectivity regression. We release the pretrained checkpoint, benchmark datasets, and fine-tuning code to support reproducible lunar AI research.

---

387. Conflict-Predictive Variable Horizons in Multi-Drone Distributed Model Predictive Control

• Authors: Linda Mümken , Michael Schwung , Stefan Lier , Andreas Schwung
• URL: https://arxiv.org/abs/2609.13270
• Abstract:

  In distributed model predictive control for multi-drone collision avoidance, a fixed prediction horizon forces a compromise: a short horizon is inexpensive but reacts late to approaching neighbors, whereas a long one anticipates conflicts at a per-step cost that grows superlinearly with its length. We propose a conflict-predictive variable horizon that each drone sets locally, leaving the distributed model predictive control itself unchanged. From a short history of observed positions, a drone extrapolates the flight lines of its neighbors, tests each against its own using confidence funnels that narrow with prediction range, and obtains each time to conflict in closed form. The horizon is then the smallest admissible value whose planning window covers the farthest predicted conflict. It collapses to its minimum in clear airspace and grows only when a conflict lies ahead. Provided this minimum meets a single computable feasibility bound, we prove that recursive feasibility and asymptotic stability are preserved for every horizon the policy can select. These guarantees hold for a linear model, and a cascaded inner loop reduces each quadrotor’s translational dynamics to a perturbed double integrator, so they carry over to the linearized quadrotor model and, as practical stability, to the full nonlinear one. In simulation on dense antipodal-swap benchmarks, the variable horizon reduces both per-step solver cost and total computation well below those of a long fixed horizon, and it maintains separation in every run, which a short fixed horizon of comparable per-step cost does not.

---

388. Forward-Facing Near-Infrared Adds Little to Colour for Farm-Machinery Traversability: A Site-Disjoint Evaluation of Sensor-Dependent Spatial Leakage

• Authors: Sungwoo Kang
• URL: https://arxiv.org/abs/2609.13265
• Abstract:

  Near-infrared (NIR) imaging does not consistently outperform standard color cameras for daytime agricultural traversability once spatial data leakage is eliminated. Prior benchmarks suggesting an NIR advantage used sequence-level splits that permitted spatially autocorrelated imagery into test sets, artificially inflating NIR performance, especially on difficult paddy-boundary segmentation. When evaluated across strictly held-out recording sites using the AI Hub autonomous driving corpus, none of the four tested configurations (color, NIR, a luminance control, or their fusion) reliably surpasses standard color across any of the five traversability classes. Furthermore, apparent performance gains observed at a single held-out site completely fail to replicate when the held-out site is rotated, demonstrating that the observed benefits were site-specific artifacts rather than generalizable improvements. Correcting for spatial leakage erases the apparent lead of NIR entirely rather than uniformly degrading performance across all sensor types. Consequently, adding a forward-facing NIR camera to a daylight agricultural vehicle sensor suite remains unproven, and evaluations compromised by location leakage risk distorting sensor rankings and misleading procurement decisions.

---

389. BEACON: Behavior and Appearance Control for Subject-Specific Video Generation

• Authors: Pokrzywa Baptiste , Nabyl Quignon , Yara Bahram , Muhammad Osama Zeeshan , Antitza Dantcheva , Eric Granger
• URL: https://arxiv.org/abs/2609.13264
• Abstract:

  Generating human-centric videos that preserve both visual identity and person-specific expressive behavior remains a fundamental challenge. In addition to reproducing appearance, a model must replicate the facial behaviors that characterize how a subject expresses emotion over time. However, most state-of-the-art methods condition generation on a single reference image, which contains no information about these temporal dynamics. As a result, they tend to preserve the subject’s visual identity but often produce expressions with limited variation and weak subject specificity. To mitigate this issue, we introduce BEACON, a lightweight framework for person-specific video generation that produces more expressive videos by disentangling visual identity from expressive behavior. BEACON conditions generation on two complementary signals: a reference image encoding the identity and a reference video capturing subject-specific facial dynamics. By conditioning on these complementary signals, BEACON generates videos that better preserve both the subject’s appearance and characteristic facial dynamics, while also supporting identity-expression transfer. Our experiments on the MEAD and RAVDESS datasets show that by fine-tuning on approximately 2,000 pairs and updating about 1% of the pretrained Wan video diffusion model, BEACON improves facial expressivity over state-of-the-art video generation methods while maintaining competitive identity preservation.

---

390. From Process Loss to Assembly Bonus: Human-Grounded Diagnosis of Multi-Agent LLM Collaboration

• Authors: Ala N. Tak , Teruhisa Misu , Kumar Akash , Zhaobo K. Zheng , Kevin H. Joo , Jonathan Gratch
• URL: https://arxiv.org/abs/2609.13261
• Abstract:

  LLM agents are increasingly used for collaborative problem solving and human-group simulation. This makes outcome-only evaluation insufficient: if LLM groups are used as models of human groups, we need to know whether they succeed or fail through human-like deliberative mechanisms. We compare human group chats with matched LLM deliberation traces on Wason-style deductive reasoning, then test whether the same process signatures generalize to analogical, abductive, and analytical tasks. Humans and LLMs show the same assembly bonus asymmetry: discussion improves the average member more often than the best initial member. Initial-answer diversity accounts for the effect of model heterogeneity, increasing movement in both corrective and destructive directions. The main differences are process-level. Compared with humans, LLM groups follow majorities more often, surface less unique information, and converge earlier; correct minority signals succeed mainly when re-expressed early. Interventions motivated by human group-decision research yield modest improvements in collective outcomes, but do not remove the coordination bottleneck. Together, these results suggest that LLM groups can reproduce some outcome-level patterns of human deliberation while diverging in the mechanisms that generate assembly bonus and process loss, with implications for group simulation and human-AI collaboration.

---

391. TryOnReward: Learning Foveated Consistency for Reinforcement Fine-Tuning of Virtual Try-On

• Authors: Xueheng Li , Yong Liu , Xiaolong Fu , Wen Xue , Chengjun Xie , Yipeng Sun , Yan Li , Simiu Gu
• URL: https://arxiv.org/abs/2609.13259
• Abstract:

  Virtual Try-On (VTON) aims to dress a person with the reference garment, producing visually reasonable results aligned with human preferences. Turning this preference-oriented goal into an actionable objective relies on a scoring function aligned with human taste. However, classic fidelity metrics exhibit weak correlation with human judgments, and generic VLMs fail to provide the discriminative granularity demanded by try-on quality evaluation, which hinges on faithfully preserving garment and person details. This shortcoming is further exacerbated in the reinforcement fine-tuning (RFT) optimization and leads to severe reward hacking. To this end, we present TryOnReward, a fine-grained reward model tailored for VTON. Built on a vision-language backbone, it adopts a foveation calibration objective that grounds each quality dimension in the relevant region to avoid global shortcut learning. Meanwhile, TryOnReward jointly optimizes pairwise preferences and per-dimension quality scores via margin-aware supervision, leveraging both relative and absolute quality signals. For model training and evaluation, we build TryOnReward-100K, a human-annotated per-dimension rating dataset, alongside TryOn-Bench and TryOnRewardBench, two benchmarks covering diverse real scenarios. Extensive experiments confirm that TryOnReward significantly outperforms generic judges in human preference alignment, and when serving as the RFT reward function, it consistently yields human-preferred try-on results across multiple baselines.

---

392. Sampling headroom is not selection gain: a compute-value audit of test-time scaling for video world models

• Authors: Yuhua Jiang , Junjie Lu , Feifei Gao
• URL: https://arxiv.org/abs/2609.13257
• Abstract:

  Test-time scaling (TTS) can improve generation only when additional compute produces better candidates and the system can reliably identify them. This distinction is especially important for video world models, where a wider sample pool may contain stronger rollouts without improving the output that is ultimately selected. We introduce the Compute-Value Audit (CVA), a sequential framework that asks whether extra sampling creates opportunity, observable signals provide a reliable state, that state supports a beneficial action, and the resulting gain exceeds the full entry fee of generation and verification. On 192 Physics-IQ scenes, expanding the pool from 4 to 16 candidates increases oracle quality by +9.23 IQ (95% CI [+7.44, +11.14]), but Flow, Cycle, and VideoReward fail to recover this headroom reliably. Across three generators, none of twelve adaptive-depth policies outperforms uniform compute; they recover only 42-69% of the measured entry fee. A matched-60-NFE Predict-and-Perturb intervention on VideoPhy2 is likewise negative across three fresh-seed replicas. These negative results are not universal: anchor-explorer passes all four stages in a sparse PRM800K setting, MMLU-Pro exposes the gap between predictive state and useful action, and a privileged paired future establishes a positive video upper bound. Together, these results show that sampling headroom has deployment value only when it can be converted into a reliable decision whose benefit survives the complete compute charge. Code is available at this https URL .

---

393. (How) Do MLLMs Report Bistable Images Like Humans?

• Authors: Ryota Takatsuki , Tomoki Doi , Amane Watahiki , Anil K. Seth , Hitomi Yanaka
• URL: https://arxiv.org/abs/2609.13254
• Abstract:

  Bistable images such as the duck-rabbit are classic stimuli in which one image supports multiple mutually incompatible interpretations, typically reported one at a time in humans. We ask whether multimodal large language models (MLLMs) show similar report behavior and what internal computations support it. Using the LLaVA family, we study two tractable dimensions: modulability, whether reports can be biased by bottom-up visual cues and top-down linguistic priors, and exclusivity, whether responses commit to a single interpretation. We test both on the canonical duck-rabbit and on synthetic Visual Anagrams to mitigate memorization confounds. Behaviorally, both visual and linguistic manipulations systematically shift reports in human-consistent ways, while responses remain predominantly exclusive. Mechanistically, these effects arise from competing image-token representations, distinct pathways for bottom-up and top-down modulation, and a link between exclusive reporting and object-count encoding. Code and data are available at this https URL .

---

394. An Evolutionary Computation Framework for Multi-Agent Q-Learning with Mean-Field Environmental Feedback

• Authors: Lichen Wang , Shijia Hua , Linjie Liu
• URL: https://arxiv.org/abs/2609.13253
• Abstract:

  Multi-agent reinforcement learning in networked populations is governed by the interaction between individual adaptation, local encounters, and changing environmental conditions. To study this interaction, we formulate a coupled learning–environment model in which agents update stateless $Q$-values on a fixed graph, while their population-average behavior drives an environmental variable that dynamically modifies the payoff matrix. Under a first-order mean-field closure, we derive a deterministic transport equation for the population distribution of $Q$-values and couple it with a projected discrete update for the environmental state. The resulting model is evaluated against finite-network Monte Carlo simulations on random regular, Erdős–Rényi, Barabási–Albert, and random geometric graphs. Across the tested parameter ranges, the mean-field system reproduces the main macroscopic cooperation and environmental trajectories, and the trajectory-level root-mean-square error generally decreases with population size and average degree. The analysis further shows that environmental feedback reshapes the learned action-value ordering, while reinforcing feedback can produce pronounced dependence on the initial learning bias and resource level. The environmental timescale also plays an important role: a rapid response can drive the resource state to a boundary before learning adapts, whereas a slower response preserves the interaction between behavioral learning and environmental recovery. These results provide a population-level description of coupled reinforcement learning and environmental dynamics and characterize the performance of the mean-field approximation within the tested network and parameter ranges.

---

395. Evaluation of MLLM-Agnostic Plug-and-Play Keyframe Selection Methods for Long Video Understanding

• Authors: Dilip Sarkar , Md. Safayet Islam , Liang Liang
• URL: https://arxiv.org/abs/2609.13250
• Abstract:

  Multimodal large language models (MLLMs) cannot process every frame of a long video because of limitations in visual-token and computational budgets. Three primary approaches have been proposed to enhance their long-video understanding capabilities: (i) Retraining an MLLM on a large video corpus and/or extending its input length; (ii) Training an adapter for a specific MLLM that takes the entire video and the query as input and selects the most relevant video frames; and (iii) Developing a training-free, plug-and-play (PaP) adapter that is MLLM-agnostic. We refer to the third approach as PaP keyframe selection. A PaP method may use only candidate video frames without considering the query, or it may use both candidate video frames and the query. The first approach is prohibitively expensive. The second approach requires substantial training time and computational resources, but it is accessible to many because an adapter contains significantly fewer trainable parameters than an entire MLLM. The third approach has the lowest computational cost and is therefore broadly accessible. To the best of our knowledge, only five PaP methods have been reported within the past year. All of these methods have been evaluated on one or more video question-answering benchmarks and have demonstrated improvements in long-video understanding. However, the methods were evaluated on different benchmarks using different MLLMs. We present a comprehensive evaluation of these five methods using three MLLMs across three long-video understanding benchmarks. Our results show that QAaF achieves the best performance in 13 of the 15 aggregate evaluation settings, while FOCUS ranks second overall. These results provide a common experimental reference for comparing training-free keyframe-selection methods for MLLMs.

---

396. ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action Models

• Authors: Manan Tayal , Akshay Nambi
• URL: https://arxiv.org/abs/2609.13231
• Abstract:

  Vision-Language-Action (VLA) models demonstrate strong generalization in robotic manipulation and navigation, but existing fine-tuning methods provide limited safety guarantees. Current approaches primarily rely on Lagrangian optimization that enforces safety through soft penalties on expected cumulative cost, often resulting in residual constraint violations or overly conservative behavior. Moreover, learning safety in visual domains is challenging due to the absence of dense per-step safety annotations. We propose ShieldVLA, a safety-aligned fine-tuning framework for VLA models based on Hamilton-Jacobi (HJ) reachability. ShieldVLA learns a model-free approximation of the HJ reachability value function directly from visual observations to estimate the safe operating region. The learned safety critic gates policy optimization by separating reward maximization within feasible regions from recovery near unsafe states, avoiding persistent reward-cost trade-offs. To enable scalable supervision in visual environments, we introduce rubric-based VLM safety scores that convert semantic safety feedback into structured critic targets without requiring manual cost labels. Across five navigation and manipulation benchmarks spanning multiple VLA backbones, ShieldVLA reduces cumulative safety cost by 57% on average and improves task success rate by +0.13 over SafeVLA.

---

397. Chemical and geometric representation fidelity improves drug–target affinity prediction

• Authors: Yixiao Li , Yining Qian , Yefan Chen , Zenghui Chen , Jiayue Sun , Yuhai Zhao , Cheng Tan , An-Yang Lu
• URL: https://arxiv.org/abs/2609.13230
• Abstract:

  Predicting drug–target binding affinity (DTA) requires models to distinguish subtle chemical and structural determinants underlying molecular recognition. Although recent approaches increasingly incorporate richer drug and protein information, such information may be compressed, homogenized or discretized during representation construction, causing affinity-relevant distinctions to be lost before interaction modelling. We hypothesized that this representation-stage information loss constitutes an upstream bottleneck that cannot be reliably overcome by increasingly complex interaction predictors. To test this hypothesis, we developed ReGeoDTA, a representation-preserving framework that maintains affinity-relevant chemical heterogeneity in molecular representations and continuous geometric relationships in protein structures. Across three benchmark datasets, ReGeoDTA consistently improved affinity prediction, and the proposed representation-preserving strategies retained their benefits across diverse DTA architectures. Controlled representation degradation progressively reduced predictive performance, whereas increasing downstream predictor complexity failed to recover information lost during representation construction. These findings identify representation fidelity as an upstream design principle for accurate and generalizable drug–target affinity prediction, with potential implications for computational compound prioritization.

---

398. Planning as Dynamics Relaxation: Hippocampal Recurrent Network Realizes Optimal Goal-Directed Navigation

• Authors: Yuhang He , Junfeng Zuo , Tianhao Chu , Si Wu
• URL: https://arxiv.org/abs/2609.13219
• Abstract:

  Neural correlates of spatial cognitive map are well documented, yet exactly how neural circuits perform spatial navigation in complex environments - e.g., reaching a goal while avoiding obstacles - remains largely unclear. Here, we show that a hippocampal network with appropriate recurrent connections can naturally achieve optimal goal-directed navigation via its relaxation dynamics. Specifically, we consider that the recurrent weights between the neurons represent the transition probabilities between spatial locations encoded by neurons; obstacles such as walls and blocked corridors are therefore reflected by the vanishing of connection weights. This connection pattern can be learned in the hippocampus via behavioral-timescale synaptic plasticity (BTSP) while the animal is exploring the environment. When a goal signal is presented, the network dynamics will relax into an activity field representing the goal location. We prove that this field is mathematically equivalent to the desirability field of a Linearly-solvable Markov Decision Process (LMDP), and the local log-gradient of the field indicates the navigation direction. Both theoretical analyses and simulations demonstrate that this recurrent network dynamics-mediated navigation is efficient and robust in environments with complex obstacle layouts. Moreover, only low-rank updates of the network’s connection pattern are needed when the environment has local changes. We hope this study offers insight into a general circuit principle for planning in abstract rational maps in the brain beyond spatial navigation.

---

399. Diagnosing Faults in Reinforcement Learning Simulators and World Models with Canonical Polynomial Invariants

• Authors: Tesfay Zemuy Gebrekidan , Hadush Hailu Gebrerufael
• URL: https://arxiv.org/abs/2609.13194
• Abstract:

  A large literature builds physical structure into learned dynamics on the premise that models respecting the underlying physics predict better. We test that premise using exact polynomial invariants recovered from trajectories and canonicalised as reduced Gröbner bases over $\mathbb{Q}$. On Acrobot, exactness provides little benefit for prediction: a consistency regulariser reduces algebraic residual while leaving rollout fidelity essentially unchanged, and a shaping potential recovered from a system with a 100% mass error accelerates learning as effectively as the correct potential. Exact canonical invariants instead prove valuable for diagnosis. We develop two procedures: screening, which identifies the violated physical constraint, and attribution, which recovers the faulty invariant and identifies the responsible physical parameter. To enable this, we introduce normal-form deflation and quotient-space recovery. Across fifteen injected faults, screening localises every broken constraint with no false alarms, whereas observation-space baselines do not localise any; attribution recovers the responsible parameter on all seven parameter faults. Paired difference tests detect all faults, showing that the advantage is localisation rather than detection. Perturbing reference generators by $10^{-4}$ preserves 14–15/15 localisations, showing that screening does not require exactness, whereas ideal-equality decisions distinguish perturbations of only $10^{-12}$, showing that exactness is required for algebraic comparison. Applied to 350 release pairs across eleven RL environments, the diagnostic finds no evidence of changed simulator dynamics, instead revealing properties of the benchmark implementations themselves.

---

400. LLMs or Naive Bayes? Old Gems or New Ways

• Authors: Mohammad Firas Sada , Dmitry Mishin , John Graham , Seungmin Kim , Mahidhar Tatineni , Frank Würthwein
• URL: https://arxiv.org/abs/2609.13185
• Abstract:

  Large language models (LLMs) prompt a recurring question in research computing: should classical methods like Naive Bayes (NB) be retired? We benchmark Complement Naive Bayes against zero-shot and few-shot LLMs spanning four model families and a 37x range in scale (27B to a 1T-parameter mixture-of-experts) across text classification tasks. LLMs dominate only in zero-data regimes (98.0% vs 88.2% on Amazon Polarity sentiment), and even that win is contamination-prone: on a low-contamination sentiment task NB beats the zero-shot LLM (81.7% vs 73.0%). However, once labeled data is available (e.g., AG News), NB reaches 89.1% accuracy, statistically indistinguishable from the zero-shot 27B LLM (89.0%) and better than the 397B frontier model (84.8%), at thousands of samples/sec on a commodity CPU. Fine-tuned DistilBERT reaches 90.6% but at far lower throughput than NB at batch size 1 (Table 2). Our measured GPU throughput analysis shows small-LLM batched inference is 40-486x slower than NB CPU inference (the multiplier depends strongly on the host CPU), exposing a structural gap bounded by memory bandwidth, with roughly two orders of magnitude lower energy per sample. For resource-constrained HPC practitioners performing text classification with labeled data, NB remains the optimal choice. We show the decision line is task-dependent (NB reaches LLM parity around $N \sim 10^4$ labels for topic classification, while zero-data sentiment favors the LLM at all N tested) and provide a Kubernetes Helm operator that automates model selection using configurable thresholds and verifiable Prometheus metrics.

---

401. Is Bash All You Need? An Empirical Study of Tool Interfaces for Enterprise Digital Worker Agents

• Authors: Hazel Mak , Susheel Suresh , Sahil Bhatnagar , Barry Wang , Chhaya Methani , Alejandro Gutierrez Munoz
• URL: https://arxiv.org/abs/2609.11999
• Abstract:

  In this study, we examine whether a general shell can outperform specialized tools on enterprise tasks. Shell-based agents have shown strong results in coding, but enterprise work also involves moving between applications and services, coordinating with coworkers, and performing professional analysis. We compare five tool interfaces on TheAgentCompany and APEX-Agents using Opus-4.8 and GPT-5.5: typed tools, typed tools plus bash, bash alone, bash with persistent agent-synthesized tools, and programmatic tool calling (PTC), which runs programs whose actions are restricted to a typed tool catalog. Bash alone outperforms typed tools on both benchmarks, improving score by 21.8-24.5 pp on TheAgentCompany and 4.8-7.4 pp on APEX-Agents while using 19-72% fewer total tokens. Adding typed tools or persistent tool synthesis to bash produces no detectable pooled score gain. PTC uses fewer tokens than direct typed calls with broadly similar task performance, but generally underperforms bash alone in both quality and cost efficiency. For enterprise practitioners, these results favor bash alone when arbitrary execution can be isolated and PTC when security or compliance policies require a fixed tool catalog.

---

402. From Semantic to Token Communication: The Next Paradigm for Large-Model-Driven 6G Intelligent Connectivity

• Authors: Yu Ma , Zhen Gao , Li Qiao , Xiaoyuan Zhang , Mahdi Boloursaz Mashhadi , Yin Xu , Wenjun Xu , Xiaodong Xu , Kaibin Huang , Jiangzhou Wang , Rahim Tafazolli , Sheng Chen , Tony Q. S. Quek , Ping Zhang
• URL: https://arxiv.org/abs/2609.10714
• Abstract:

  The ambitious requirements of sixth-generation (6G) networks are driving communication systems from reliable bit delivery toward meaning-aware and task-oriented connectivity. Large models (LMs), with strong multimodal understanding and generation capabilities, have accelerated this shift and made semantic communication (SemCom) increasingly practical. Yet current LM-driven SemCom remains fragmented: semantic representations are typically tied to specific modalities, models, or tasks. While the bit provides a universal unit for digital transport, there is still no analogous unit for representing and processing semantics, which limits interoperability, theoretical unification, and scalable system design. We argue that tokens provide a natural candidate for this missing abstraction. Two trends support this: unified multimodal LMs now encode text, images, audio, video, and robot actions in one token space, while distributed LM inference already generates substantial token-level traffic through expert routing, cache transfer, and speculative decoding. Token communication (TokenCom) emerges by unifying these trends, using the LM’s native processing unit as a communication abstraction above the bit level and enabling importance assignment, error handling, and resource allocation directly at token granularity. This survey traces the evolution from LM-driven SemCom to TokenCom. We review three major directions of LM-driven SemCom: source-centric semantic coding, channel semantics for physical-layer tasks, and collaborative edge-device intelligence. We then examine the token abstraction, the transmission techniques it requires, and two emerging paradigms, namely TokenCom for LM services and for embodied and agentic intelligence. Finally, we identify open challenges toward unified, scalable, and AI-native 6G communication systems.

---

403. Multilingual Agent System for Inclusive Wildfire Evacuation Guidance

• Authors: Shruti Kulkarni , Lynn Tong , Aditi Namboodiripad , Chelyah Miller , Helen Lin , Peeyush Patel , Bogdan Bistriceanu , Diane Myung-kyung Woodbridge
• URL: https://arxiv.org/abs/2609.03301
• Abstract:

  Wildfire seasons have become 84 days longer in the current days than in the 1970s, causing enormous threats to one’s financial status and short- and long-term health. During the fire, public agencies send out emergency messages to provide warnings and orders. Although 26 million people in the US have limited English proficiency, over 80% of those messages are only delivered in English, which can cause disproportionate information distribution and awareness. In order to better serve marginalized communities during emergencies, the authors developed BEACON, a service that provides comprehensive and personalized evacuation guidance, including navigation routes, personalized checklists, and a chatbot in the language that a user uses. Our current system ingests data including fire perimeter information, evacuation order status, and shelter information from Watch Duty. When a user is within a certain proximity from the fire, the system utilizes real-time GPS locations and nearby weather data from the National Oceanic and Atmospheric Administration (NOAA) to predict fire danger levels. The assessment model refreshment are dynamically scheduled based on fire progress and trends using XGBoost. If the location has a likelihood of fire danger, the system sends alerts with evacuation routes outputted from a polygon-avoidant routing pipeline. The application provides a context-aware multilingual agent that users can communicate with and is tightly connected to other features of the application. In addition, based on data that the user entered, the system dynamically generates and checks off personalized reminder items to provide an organized evacuation plan. The system’s user interface dynamically changes its language settings based on the language the user most recently used in either setting or chatbot conversation for all the application elements.

---

404. Natural-Language to SysMLv2 Translation via Conformance-Driven Iterative Refinement

• Authors: Chance LaVoie , Eladio Andujar Lugo , Taylan G. Topcu , Levent Burak Kara
• URL: https://arxiv.org/abs/2607.14162
• Abstract:

  Model-Based Systems Engineering (MBSE) relies on formal system models as primary technical artifacts for representing requirements, structure, and behavior across the system lifecycle. With the standardization of SysMLv2 as a textual language, interest is increasing in translating natural-language descriptions directly into executable models. For practical deployment, generated models must be accepted by industrial modeling environments, not merely satisfy grammar constraints. We present a conformance-checker-driven framework for reliable natural-language-to-SysMLv2 translation that enforces production-level acceptance as the termination condition. The system embeds a SysMLv2 conformance checker within a generate-check-repair loop. Each model is evaluated using the checker, and deterministic diagnostics are incorporated into revisions until zero conformance errors are achieved. Using the production checker as the oracle ensures the framework targets deployability rather than grammar plausibility. We evaluate the approach on the full SysMBench prompt set of 151 prompts across four large language model backends, yielding 604 prompt-model cases. Single-shot generation achieves 51.16% production-conformance acceptance, while our approach achieves 100.00% conformance. By elevating production conformance from a post-processing check to a control mechanism within generation, the framework converts probabilistic outputs into production-accepted SysMLv2 artifacts suitable for loading, visualization, and engineering use.

---

405. Physically Aware Radiomics Without Interpolation: Disentangling Voxel Geometry and Signal Modification in CT and MRI

• Authors: David Corral Fontecha , Juan Miranda Bautista , Pablo Menendez Fernández-Miranda , Sergio Rubio-Martín , Lara Lloret Iglesias , Jose A. Vega
• URL: https://arxiv.org/abs/2607.12399
• Abstract:

  Objective: Radiomic texture features are usually computed in voxel-index neighborhoods, implicitly assuming isotropic spatial relationships. In anisotropic images, this can confound voxel geometry with interpolation-induced signal changes. We developed a voxel-spacing-aware radiomic framework that incorporates physical geometry into texture computation without resampling. Approach: We modified PyRadiomics to account for voxel spacing while preserving the native image signal. Four configurations were compared: native non-resampled extraction (NR), isotropic resampling (RS), voxel-spacing-aware extraction (VS), and fake-isotropic preprocessing (FK), in which spacing metadata were overwritten without altering the image array. Experiments included 685 LIDC-IDRI pulmonary nodules and 209 I-SPY2 breast MRI cases, with 196 radiomic descriptors. Robustness was assessed using ICC, within-subject variability, Friedman testing, feature selection, machine learning, a multilayer perceptron, and external validation. Main results: VS showed near-native agreement with NR: median ICC(A,1) was 0.9976 in CT and 0.9984 in MRI. RS produced lower agreement and larger deviations, while FK showed intermediate behavior, confirming that spacing metadata alone can affect radiomic features. Gradient-derived and neighborhood-sensitive descriptors were most affected by preprocessing. VS preserved predictive performance comparable to NR in external CT validation, whereas MRI showed greater variability across preprocessing strategies and classifiers. Significance: Voxel-spacing-aware extraction separates geometric modeling from interpolation-induced signal modification while preserving the native image signal, offering a coherent alternative to isotropic resampling for radiomic analysis of anisotropic CT and MRI.

---

406. Generalization Can Emerge in Tabular Foundation Models From a Single Table

• Authors: Junwei Ma , Nour Shaheen , Alex Labach , Amine Mhedhbi , Frank Hutter , Anthony L. Caterini , Valentin Thomas
• URL: https://arxiv.org/abs/2511.09665
• Abstract:

  Deep tabular modelling increasingly relies on in-context learning where, during inference, a model receives a set of $(x,y)$ pairs as context and predicts labels for new inputs without weight updates. We challenge the prevailing view that broad generalization here requires pre-training on large synthetic corpora (e.g., TabPFN priors) or a large collection of real data (e.g., TabDPT training datasets), discovering that a relatively small amount of data suffices for generalization. We find that simple self-supervised pre-training on just a \emph{single} real table can produce surprisingly strong transfer across heterogeneous benchmarks. By systematically pre-training and evaluating on many diverse datasets, we analyze what aspects of the data are most important for building a Tabular Foundation Model (TFM) generalizing across domains. We then connect this to the pre-training procedure shared by most TFMs and show that the number and quality of \emph{tasks} one can construct from a dataset is key to downstream performance.

---

407. Beyond Quacking: Deep Integration of Language Models and RAG into DuckDB

• Authors: Anas Dorbani , Sunny Yasser , Jimmy Lin , Amine Mhedhbi
• URL: https://arxiv.org/abs/2504.01157
• Abstract:

  Knowledge-intensive analytical applications retrieve context from both structured tabular data and unstructured, text-free documents for effective decision-making. Large language models (LLMs) have made it significantly easier to prototype such retrieval and reasoning data pipelines. However, implementing these pipelines efficiently still demands significant effort and has several challenges. This often involves orchestrating heterogeneous data systems, managing data movement, and handling low-level implementation details, e.g., LLM context management. To address these challenges, we introduce FlockMTL: an extension for DBMSs that deeply integrates LLM capabilities and retrieval-augmented generation (RAG). FlockMTL includes model-driven scalar and aggregate functions, enabling chained predictions through tuple-level mappings and reductions. Drawing inspiration from the relational model, FlockMTL incorporates: (i) cost-based optimizations, which seamlessly apply techniques such as batching and caching; and (ii) resource independence, enabled through novel SQL DDL abstractions: PROMPT and MODEL, introduced as first-class schema objects alongside TABLE. FlockMTL streamlines the development of knowledge-intensive analytical applications, and its optimizations ease the implementation burden.

---

408. Towards Optimizing SQL Generation via LLM Routing

• Authors: Mohammadhossein Malekpour , Nour Shaheen , Foutse Khomh , Amine Mhedhbi
• URL: https://arxiv.org/abs/2411.04319
• Abstract:

  Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capable large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.

---