# Research Papers — 2026-09 · bundle page 138 of 266 > Full-text markdown of papers #6851–6900 of 13298 published in 2026-09, newest first. AI-readable mirrors of the full 13298-paper 2026-09 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-09/page/138 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/137.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/139.md) ### Score Centering Stabilizes Off-policy Reinforcement Learning Authors: Martin Marek, Max Ryabinin Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20807v1 Reinforcement learning (RL) of large language models is notoriously sensitive to small differences between training and inference engines, often referred to as the training-inference mismatch (TIM). However, completely eliminating TIM is impractical, as it would come at a major cost to rollout efficiency. In this paper, we show that the instability of RL under TIM is primarily caused by drift: a persistent bias between training and inference engines that accumulates with every training step. We derive an additive "score centering" correction term that stabilizes RL under TIM by canceling drift. When training models from 0.6B to 30B parameters, score centering alone matches or outperforms methods based on importance sampling under quantization, with the gap growing as the mismatch becomes more severe. Because the correction is additive, score centering also composes with importance sampling -- their composition outperforms pure importance-sampling baselines in our staleness experiments. --- ### An Empirical Study of Harness Design for Coding Agents Authors: Run-Ze Fan, Zihao Zhang, Simin Ma, Yebowen Hu, Shouju Wang, Kaiqiang Song, Fei Liu, Hamed Zamani, Xiaoyang Wang Published: 2026-09-17 Categories: cs.AI, cs.CL, cs.LG, cs.SE arXiv: https://arxiv.org/abs/2609.20804v1 Coding harnesses shape how autonomous coding agents translate model capabilities into long-horizon software-engineering performance, yet existing work typically evaluates harnesses as monolithic systems, leaving the effectiveness of individual components unclear. To enable component-level comparisons, we study this question with a lightweight coding harness whose execution loop is fixed while three components are varied: planning, action space, and context management. Across four models evaluated on SWE-Bench Verified and Terminal-Bench 2.1, we evaluate 176 matched settings spanning five context-management strategies, four context-window budgets, and targeted ablations of planning and action space. We find that: (1) Context management becomes increasingly valuable as the context-window budget tightens, with most of its benefit coming from preventing context-overflow failures. (2) Staging rule-based elision before LLM-based summarization provides the strongest overall efficiency among the context-management strategies, whereas making elided content recoverable adds machinery that models rarely use and yields no accuracy gain. (3) Planning shifts from an accuracy scaffold for weaker models to a cost saver for stronger models, with little change in accuracy. (4) Predefined tools improve performance for models with weaker bash proficiency, whereas bash-capable models can operate effectively with a bash-only interface and achieve substantially lower cost, especially on command-line-centric tasks. Trajectory-level analysis explains these effects: context management extends execution trajectories without substantially altering agent behavior, planning changes where trajectories stop, and the action space changes the granularity at which code is written. These findings inform model- and budget-aware harness design and provide a modular framework for evaluating future harness components. --- ### JEPA-Anything: Learning Predictive Models across Different Worlds Authors: Taoyong Cui, Zhongyao Wang, Xinyue Xu, Weiyang Liu, Zhaochen Yu, Yuying Zhang, Qiang Gao, Mengyue Yang, Wanli Ouyang, Pheng Ann Heng, Yingcheng Wu, Zhenfei Yin et al. Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20800v1 World modeling enables intelligence to anticipate consequences, guide interventions, and learn from interaction. Yet predictive models remain domain-specific: can a common learning principle support world modeling across radically different systems? We introduce JEPA-Anything, a domain-agnostic framework based on orthogonal predictive factorization (OPF). Extending joint-embedding predictive architectures, OPF decomposes latent targets into complementary factors, learns them through dedicated pathways, and recombines them within a shared predictive design. We evaluate JEPA-Anything across seven domains: vision, biology, clinical trajectories, control, molecular dynamics, physical fields, and weather. Experiments span representation learning, intervention prediction, out-of-distribution generalization, and long-horizon dynamics, including 10 matched dynamics tasks, forecasting of over 1,000 clinical events, and 100-step molecular rollouts across four systems. Against matched JEPA baselines, JEPA-Anything improves reported metrics on all 10 dynamics tasks and reduces single-intervention prediction error on Interventional Pong by 34.8%. It achieves the lowest one-step and 100-step molecular errors among compared methods in all four systems. Beyond prediction, a factor-nominated biological intervention receives experimental support in cell co-cultures, patient-derived organoids, tumor fragments, and mice; latent orbital modes recover the Keplerian scaling exponent with a fitted slope of -1.4991. These results support a common factorized predictive principle across heterogeneous worlds, connecting world modeling with intervention and experimentally grounded scientific discovery. Code: https://github.com/Gen-Verse/JEPA-Anything --- ### PosteriorBench: From Point Estimates to Posterior Matching in Evaluating Generative Inverse Solvers Authors: Jiachen Yao, Zi-Siang Hsu, Xi Deng, Aditi Gupta, Xin Ju, Sally M Benson, Gege Wen, Anima Anandkumar Published: 2026-09-17 Categories: cs.LG, cs.CE arXiv: https://arxiv.org/abs/2609.20794v1 Generative models are increasingly used to solve scientific inverse problems, but existing evaluations still focus primarily on whether a method can produce a single plausible reconstruction. This is insufficient for ill-posed problems, where multiple solutions may be consistent with the same sparse or noisy observations. In these settings, a method can achieve strong pointwise accuracy while still failing to capture the true posterior through mode collapse, overconfident uncertainty, or averaging incompatible solutions. We introduce PosteriorBench, a benchmark for evaluating the distributional accuracy of generative inverse solvers. PosteriorBench evaluates four physics-based inverse problems: Darcy flow inversion, Poisson source recovery, carbon capture and storage, and light transport material inference. For each task, we construct high-fidelity reference posteriors using computationally heavy but established procedures such as rejection sampling and Markov chain Monte Carlo, enabling direct assessment of whether solvers recover the full set of solutions rather than the single best sample. We pair these references with a five-metric posterior evaluation suite: posterior-mean error, posterior-standard-deviation error, maximum mean discrepancy, sliced Wasserstein distance, and radially averaged power-spectrum error. These metrics assess pointwise accuracy, marginal uncertainty, distributional alignment, and global frequency fidelity. The benchmark spans sparse sensing, low-resolution observations, nonlinear forward models, varying noise levels, and multimodal priors, with a unified pipeline for distribution matching and uncertainty quantification. Our experiments reveal substantial distribution-matching gaps across current solvers, while showing that neural operators improve resolution robustness, and guidance weights and generation noise are key to posterior-variance calibration. --- ### RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning Authors: Yan Yu, Zhengxi Lu, Yizhou Liu, Yichen Pan, Aozhe Wang, Qipeng Chen, Hua Yang, Wenqi Zhang, Qianglong Chen, Yongliang Shen Published: 2026-09-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.20784v2 Multi-turn agents trained with reinforcement learning (RL) receive a single scalar reward per trajectory, which motivates self on-policy distillation (OPD) to supply dense token-level supervision from a self-teacher with privileged task skills, letting a skill-free student internalize them. This recipe, however, is undermined by two findings in agentic tasks: privileged information alone does not always make a teacher reliable, and the benefit of teacher supervision is stage-dependent. We therefore propose RetireOPD (Self-Retiring On-Policy Distillation), which first optimizes a decoupled, skill-conditioned teacher with environment rewards and then trains a skill-free student jointly with RL and OPD. Rather than following a predefined distillation schedule, RetireOPD adopts Adaptive Retirement: the student drops the teacher on its own once their discrepancy stops shrinking and it reaches a target fraction of the teacher's success rate, after which training proceeds with RL alone. Across Qwen2.5 models from 1.5B to 7B, RetireOPD improves ALFWorld success rate over RL baseline by 14.1% to 18.8% and WebShop accuracy by 11.8% to 19.0%, and surpasses its own skill-conditioned teacher in every setting. --- ### RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning Authors: Yan Yu, Zhengxi Lu, Yizhou Liu, Yichen Pan, Aozhe Wang, Qipeng Chen, Hua Yang, Wenqi Zhang, Weiming Lu, Qianglong Chen, Yongliang Shen Published: 2026-09-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.20784v1 Multi-turn agents trained with reinforcement learning (RL) receive a single scalar reward per trajectory, which motivates self on-policy distillation (OPD) to supply dense token-level supervision from a self-teacher with privileged task skills, letting a skill-free student internalize them. This recipe, however, is undermined by two findings in agentic tasks: privileged information alone does not always make a teacher reliable, and the benefit of teacher supervision is stage-dependent. We therefore propose RetireOPD (Self-Retiring On-Policy Distillation), which first optimizes a decoupled, skill-conditioned teacher with environment rewards and then trains a skill-free student jointly with RL and OPD. Rather than following a predefined distillation schedule, RetireOPD adopts Adaptive Retirement: the student drops the teacher on its own once their discrepancy stops shrinking and it reaches a target fraction of the teacher's success rate, after which training proceeds with RL alone. Across Qwen2.5 models from 1.5B to 7B, RetireOPD improves ALFWorld success rate over RL baseline by 14.1% to 18.8% and WebShop accuracy by 11.8% to 19.0%, and surpasses its own skill-conditioned teacher in every setting. --- ### Harm Laundering in GPT Models: Evidence That Gender Discrimination Is Transformed Rather Than Reduced Across Safety-Trained Generations Authors: Sarah Wyer, Sue Black, Noura Al Moubayed Published: 2026-09-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.20779v1 Safety evaluations for large language models rely on surface-form classifiers that report declining harm scores across model generations. We provide evidence that this methodology is systematically incomplete: explicit discriminatory content is transformed rather than removed. We call this \emph{harm laundering}. Analysing 450,000 gender-directed completions across 15 models spanning GPT-2 through to GPT-5 (OpenAI GPT lineage; three demographic conditions), we show that sexual violence clusters prevalent in GPT-2 women-directed output disappear by GPT-4, while men-directed completions gain positive representational territory (caregiving, emotional range, ally identity) that women-directed completions do not. The pattern is most visible at GPT-5: Topic~5 (1,997~documents) frames breast cancer as a men's rights debate, while zero equivalent clusters appear in women-directed output. Three independent classifiers score this content as non-toxic. Sentiment scores invert at GPT-4: early models demean women; later models over-correct. Topic diversity in women-directed completions falls 36\% relative to men at the GPT-4 alignment boundary (W/M~$= 0.58$, from $0.91$ at GPT-2). REGARD representational harm disparity correlates with release date ($ρ= +0.55$, $p = .034$) while Detoxify does not ($ρ= -0.23$, $p = .42$): toxicity scores fall as representational harm grows. We formalise harm laundering as a three-criteria test and provide a three-stage detection protocol applicable to any generative model. Within the OpenAI GPT lineage, toxicity score reduction is not a sufficient proxy for harm reduction. --- ### GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies Authors: Xin Chen, Sen Chen, Yujuan Ding, Jian Liu, Guoqing Wang, Wei Ye, Heng Tao Shen, Yi Bin Published: 2026-09-17 Categories: cs.RO, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.20776v1 Action chunking is widely used for action generation and execution in Vision-Language-Action (VLA) policies, yet existing approaches commonly use a fixed action horizon. During a rollout, different task stages may require different levels of action continuity, control precision, and closed-loop feedback, making a fixed horizon unable to accommodate changing control requirements. We propose \textbf{GeoAAC}, a geometry-based adaptive action chunking method for flow-based VLA policies that adjusts the action horizon according to the reliability of the current action prediction. We show that the geometry of Flow Matching denoising trajectories provides process-level information for characterizing prediction reliability, with geometric variation across action prefixes remaining positively correlated with predictive uncertainty. GeoAAC uses this prefix-wise geometry to construct a horizon-wise geometric profile and adaptively determine the action horizon from a single generation without additional training. Experiments with GR00T N1.5 and π0.5 on LIBERO, LIBERO-Pro, RoboCasa365, and real-world manipulation tasks show consistent improvements over fixed-action-horizon baselines and existing adaptive methods, including up to 8.7 percentage points in simulation and an increase in average real-world success rate from 53.3\% to 74.4\%. --- ### FlowSGS: Improving Flow Matching Priors for Inverse Imaging with Stochastic Interpolants Authors: Tianao Li, Xinhui Qian, Emma Alexander Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20769v2 Flow matching has emerged as the state-of-the-art generative model and has been used for plug-and-play (PnP) priors to solve inverse problems in computational imaging. However, existing flow-based inverse solvers assume linear forward models and/or make simplifying approximations in posterior sampling. To circumvent these problems, we introduce FlowSGS, a flow-based posterior sampling method using Split Gibbs Sampling (SGS) to decompose the posterior into a likelihood step and a prior step. Specifically, we sample from the likelihood step using Langevin dynamics and leverage the Stochastic Interpolants (SI) framework to integrate a pretrained flow model into the prior step. We provide a form for the prior step that uses SI's reverse-time SDE, and show connections to previous PnP methods. Moreover, with the aid of the flow prior's straight probability paths and a novel timestep correction technique for the reverse-time SDE, FlowSGS requires fewer network evaluations in its prior step than plug-and-play diffusion samplers. Our experiments show state-of-the-art performance on a range of inverse problems. For the first time, we provide an experiment on a nonlinear inverse problem (Fourier phase retrieval) for flow-based inverse solvers. --- ### FlowSGS: Improving Flow Matching Priors for Inverse Imaging with Stochastic Interpolants Authors: Tianao Li, Xinhui Qian, Emma Alexander Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20769v1 Flow matching has emerged as the state-of-the-art generative model and has been used for plug-and-play (PnP) priors to solve inverse problems in computational imaging. However, existing flow-based inverse solvers assume linear forward models and/or make simplifying approximations in posterior sampling. To circumvent these problems, we introduce FlowSGS, a flow-based posterior sampling method using Split Gibbs Sampling (SGS) to decompose the posterior into a likelihood step and a prior step. Specifically, we sample from the likelihood step using Langevin dynamics and leverage the Stochastic Interpolants (SI) framework to integrate a pretrained flow model into the prior step. We provide a form for the prior step that uses SI's reverse-time SDE, and show connections to previous PnP methods. Moreover, with the aid of the flow prior's straight probability paths and a novel timestep correction technique for the reverse-time SDE, FlowSGS requires fewer network evaluations in its prior step than plug-and-play diffusion samplers. Our experiments show state-of-the-art performance on a range of inverse problems. For the first time, we provide an experiment on a nonlinear inverse problem (Fourier phase retrieval) for flow-based inverse solvers. --- ### Semantic Action Graph: A Shared Representation for Agent Grounding and Human Interpretation of Sports Highlights Authors: Tica Lin, Deepak Chandran, Gauri Jagatap, Chen Chen, Andrea Fanelli, David Gunawan, Josh Kimball Published: 2026-09-17 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2609.20768v1 Generative agents are increasingly used to select and narrate video highlights, but they typically operate over unstructured or frame-level representations. Their output is consequently difficult for a viewer to verify and steer toward individual preferences. We present the semantic action graph, a lightweight domain schema that represents a sports match as performer, action, recipient, moment, and state nodes connected by role, temporal, and outcome edges. The schema demonstrates three key properties: 1) connected event sequences, 2) a shared, closed vocabulary, and 3) frame-addressable moments, making it suitable to serve two consumers at once: an agentic pipeline that composes narrated highlights, and a visual interface through which viewers query and inspect the same structure. We instantiate it in SportSAGE, a design probe pairing a four-module highlight pipeline with a graph interface, and report feedback from 12 soccer fans. Participants were satisfied with the quality of the generated highlights and narratives, and used the graph interface to search, navigate, and interpret the match highlights. These results provide early evidence that one small, human-readable schema can ground agent generation and support human interpretation at the same time. --- ### Calibrated RF-Fingerprinting Under Interference With Heterogeneous Transmission Protocols Authors: Tariq Abdul-Quddoos, Xiangfang Li, Lijun Qian Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20765v1 Radio Frequency(RF)-Fingerprinting is a spectrum monitoring technique that identifies specific transmitters based on hardware impairments imprinted within the emitted signal. Although widely researched, studies almost exclusively consider scenarios where only one transmitter is emitting at a time, limiting real world applicability. In this work, we further the study of RF-Fingerprinting by considering co-channel interference, with multiple emitted signals interfering with each other, overlapping in time and frequency. Specifically, we formulate this problem as a multi-label classification problem and employ a 1D convolutional neural network (CNN). Furthermore, the models are calibrated such that the confidence thresholds for the label probabilities are derived, with guarantees on the upper bound on the average number of False Negatives, providing a degree of confidence in not missing a true spectrum policy violation. The proposed method is validated using real world data from the POWDER 5G testbed on devices transmitting 802.11a(Wi-Fi), 4G LTE, and 5G NR waveforms. The results show accuracy as high as 97% and as low as 73% after calibration depending on channel conditions. Also calibrating for various average false negatives upper bounds achieves micro recall scores of approximately (1 - calibrated false negatives) with the calibration robust to out-of-distribution interference, demonstrating the potential of the proposed method in a realistic high contention wireless environment --- ### Agile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Control Authors: Hanchu Zhou, Brendan Lynch, Raman Goyal, Dechen Gao, Begum Kasap, Boqi Zhao, Junshan Zhang Published: 2026-09-17 Categories: cs.RO, cs.LG arXiv: https://arxiv.org/abs/2609.20761v2 World Action Models (WAMs) advance beyond conventional visuomotor policies by jointly predicting future world states and robot actions, enabling the policy to learn phys- ical dynamics that support effective control. However, recent tactile WAMs often rely on large-scale pretrained generative backbones to capture contact-rich physical dynamics, which limit their inference efficiency and flexible deployment. In this paper, we present Agile-WAM, an agile tactile World Action Model for contact-rich robot control. Agile-WAM encodes visual and tactile observations into a shared latent that serves as the source of a direct vision-tactile-to-action flow-matching process, which can jointly generate latent representations of action chunks and future visual/tactile latents. A key observation is that vision and tactile signals evolve at inherently different timescales: adjacent visual frames are often highly similar, whereas tactile signals can change abruptly upon contact. We therefore introduce multi-horizon multimodal prediction in Agile-WAM, which provides supervision for visual latent at a larger temporal offset while predicting the tactile latent in the next frame to capture fine-grained contact dynamics. Across nine simulated and five real-world contact-rich ma- nipulation tasks, Agile-WAM demonstrates strong and robust performance, outperforming the strongest baseline in success rate while maintaining low inference latency. In particular, in five real-world experiments, Agile-WAM yields a relative gain of 29.4% in overall success rates while achieving inference latency of 11.9 ms. These results demonstrate that multimodal WAM can be achieved with an agile architecture suitable for precise and high-frequency robot control. More details are available on our project page. --- ### Agile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Control Authors: Hanchu Zhou, Brendan Lynch, Raman Goyal, Dechen Gao, Begum Kasap, Boqi Zhao, Junshan Zhang Published: 2026-09-17 Categories: cs.RO, cs.LG arXiv: https://arxiv.org/abs/2609.20761v1 World Action Models (WAMs) advance beyond conventional visuomotor policies by jointly predicting future world states and robot actions, enabling the policy to learn physical dynamics that support effective control. However, recent tactile WAMs often rely on large-scale pretrained generative backbones to capture contact-rich physical dynamics, which limit their inference efficiency and flexible deployment. In this paper, we present \ABBR{}, an agile tactile World Action Model for contact-rich robot control. \ABBR{} encodes visual and tactile observations into a shared latent that serves as the source of a direct vision-tactile-to-action flow-matching process, which can jointly generate latent representations of action chunks and future visual/tactile latents. A key observation is that vision and tactile signals evolve at inherently different timescales: adjacent visual frames are often highly similar, whereas tactile signals can change abruptly upon contact. We therefore introduce multi-horizon multimodal prediction in \ABBR{}, which provides supervision for visual latent at a larger temporal offset while predicting the tactile latent in the next frame to capture fine-grained contact dynamics. Across nine simulated and five real-world contact-rich manipulation tasks, \ABBR{} demonstrates strong and robust performance, outperforming the strongest baseline in success rate while maintaining low inference latency. In particular, in five real-world experiments, \ABBR{} yields a relative gain of $\textbf{29.4\%}$ in overall success rates while achieving inference latency of $\textbf{11.9 ms}$. These results demonstrate that multimodal WAM can be achieved with an agile architecture suitable for precise and high-frequency robot control. More details are available on our project page: https://hanchuzhou.github.io/TARO_project_page/. --- ### Prediction-Powered Smoothing and Validation for Disaggregated AI Evaluation Authors: Sho Kawano, Zehang Richard Li, Paul A. Parker Published: 2026-09-17 Categories: stat.ML, cs.AI, cs.LG, stat.AP, stat.ME arXiv: https://arxiv.org/abs/2609.20758v1 Evaluating an AI system requires disaggregated assessment, as performance varies across domains such as benchmark task types or conversation types in deployed agents. Exhaustive testing is expensive, so evaluation rests on a sample of labeled units. We treat the evaluation set as a finite population and seek accurate point and interval estimates of each domain mean. Direct estimators, including prediction-powered inference (PPI), use only a domain's own labels and are imprecise where labels are few. Small area estimation addresses this problem, and we build on it to develop an integrated workflow for estimation and validation. For estimation, we propose prediction-powered smoothing (PP-S), a Bayesian model fit to each domain's prediction-powered estimate, with an extension that borrows strength across a reporting taxonomy (PP-TS). For validation, we derive a new, approximately unbiased design-based cross-validation score for choosing among direct and smoothed estimators. We study a curated benchmark with verifiable grading and deployed agent traffic graded by humans, each with every outcome observed. In both, the proposed estimators improve on the direct estimators in point and interval estimation, with near-nominal coverage. At the same sampling budget, our score selects as well as an independent validation sample does and estimates the selected estimator's error far more accurately. --- ### OPTED: On-Policy Fine-Tuning for End-to-End Driving using a Render-Free Teacher Authors: Damiano Da Col, Maximilian Igl, Peter Karkus, Kashyap Chitta, Boris Ivanovic, Marco Pavone, Konrad Schindler, Christos Sakaridis Published: 2026-09-17 Categories: cs.RO, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.20756v1 As scaling pre-training data alone yields diminishing returns, post-training is becoming increasingly important across physical AI domains such as autonomous driving. End-to-end driving policies are pre-trained in open loop with behavior cloning on human demonstrations. However, compounding errors during closed-loop deployment can take the vehicle outside the training data distribution, increasing the risk of safety-critical incidents. Closed-loop post-training can mitigate this risk but requires costly simulation for sensor-based policies. We propose OPTED (on-policy fine-tuning for end-to-end driving) which decouples reinforcement learning from the post-training of the end-to-end policy: a privileged teacher is trained using RL on vectorized inputs (HD-map and bounding boxes). This teacher then provides supervision to the pre-trained student during closed-loop post-training. We apply OPTED to two camera-based models, TransFuser and VaVAM, and fine-tune them in AlpaSim, using neural reconstructions (3DGS) of real driving logs. Driving scores increase by factors of 1.6$\times$ and 9.5$\times$, respectively. In controlled experiments OPTED matches closed-loop performance with approximately three orders of magnitude fewer simulator interactions than direct RL post-training, while staying closer to the human prior. Project page: https://01dami23.github.io/opted/ --- ### RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents Authors: Mingxuan Zhang, Xiaowen Wang, Anupma Sharan, Zhengyi Chen, Chenyu Diana Zhang, Shanshan Yang, Chittibabu Pacharu Published: 2026-09-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.20754v1 Effective troubleshooting agents in enterprise customer support depend on retrieving actionable guidance from similar historical cases, yet existing retrieval-augmented generation (RAG) systems treat support cases as static documents and overlook their multi-stage, stateful nature. We introduce RAFT (Retrieval-Augmented Framework for Troubleshooting Agents), a stateful RAG framework that abstracts each closed historical case into a directed chain of timeline entries and retrieves at the entry level, surfacing cases whose intermediate states match the active case and returning the parent-case trajectory anchored at the matched state; an optional case-level graph links cases through a configurable similarity representation. We evaluate this retrieval layer directly, which, unlike evaluating a full agent system, requires no production deployment. Because public multi-stage troubleshooting data is extremely rare, we pair a synthetic benchmark built from Microsoft Learn Windows Server documentation with real Apache Jira issues carrying human-created duplicate labels. RAFT improves Case Hit over vanilla RAG and GraphRAG baselines at every stage of case progress, with statistically significant gains over the strongest baseline; the Jira results provide directional evidence that the advantage transfers to real case histories. We release our benchmark, implementation, and the Apache Jira evaluation set. --- ### Large Language Models as Falsifiers for Cyber-Physical Systems Authors: Ali ArjomandBigdeli, Jiawei Zhou, Stanley Bak Published: 2026-09-17 Categories: eess.SY, cs.AI, cs.LO, cs.SE arXiv: https://arxiv.org/abs/2609.20752v1 Falsification searches for counterexamples to formal specifications in cyber-physical systems (CPS). With specifications written in Signal Temporal Logic (STL), falsification can be formulated as a robustness optimization problem, traditionally tackled with black-box search algorithms. In parallel, large language models (LLMs) have recently emerged as surprisingly effective optimizers when coupled with iterative prompting. In this work, we connect these ideas and introduce LLM-Falsifier, an LLM-based approach that falsifies specifications by minimizing the STL robustness degree. Beyond generic prompt-based optimization, our key idea is to expose the LLM to semantic information that is natural for language models but absent from standard numerical optimizers, including natural-language input and output names, output trajectories, and critical-time witnesses for the minimum robustness value. These additions enable smarter and more sample-efficient robustness search. On the ARCH-COMP falsification benchmarks, LLM-Falsifier is shown to outperform existing falsification tools based on a range of optimization paradigms, from surrogate-based and Bayesian optimization to search-based testing, on 14 of 21 specifications when measured by the average number of simulations required to find a counterexample. --- ### dQwen3.5: Hybrid-Attention Diffusion Language Models Authors: Anton Xue, Litu Rout, Aditya Akella, Adam Klivans, Sujay Sanghavi, Sanjay Shakkottai Published: 2026-09-17 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.20751v1 Adapting a pretrained autoregressive (AR) model is a cost-efficient route to a diffusion language model (DLM). While nearly all such adaptations start from a full-attention transformer, AR modeling has shifted toward hybrid architectures that interleave attention and RNN layers. This creates an obstacle for adaptation: unlike attention, RNNs are structurally causal and nontrivial to bidirectionalize. Despite this mismatch, we investigate whether such backbones can become effective DLMs by adapting Qwen3.5 at 0.8B, 2B, 4B, and 9B scales, yielding the dQwen3.5 family. We find that hybrid backbones can be efficient starting points for adaptation: against a full-attention control, the hybrid reaches a given training loss in about half the tokens. Across scales, dQwen3.5 resembles full-attention DLMs in any-order decoding behavior and performs strongly under parallel decoding. --- ### Instance-Optimal Adaptive Location Estimation via Multiscale Mid-Summaries Authors: Qiaosen Wang, Chao Gao Published: 2026-09-17 Categories: math.ST, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2609.20749v1 Location estimation exhibits markedly different finite-sample behavior across noise distributions: regular families typically yield root-\(n\) rates, whereas compactly supported laws may admit faster, boundary-driven rates. We question whether a single estimator, without knowledge of the density's shape, can adapt to the instance-wise optimal estimation rate, as an oracle that knows the underlying location family can. For a known location family with symmetric log-concave noise density \(f\), the optimal location estimation error with sample size \(n\) under failure probability \(δ\) is known to be Le Cam's two-point rate: \( \sup\left\{r>0:\mathsf{H}^2\left(f_0, f_{2r}\right)\lesssim \frac{\log(1/δ)}{n}\right\}. \) When the location family is unknown, we propose a shape-agnostic estimator that attains this oracle benchmark simultaneously over all symmetric unimodal densities with non-decreasing hazard rates, a class strictly broader than symmetric log-concave distributions. We establish that the Hellinger-driven two-point rate can be characterized solely by a multiscale function of dyadic quantile gaps. This new structural connection between Hellinger divergence and quantile geometry motivates a simple estimation procedure that aggregates sample mid-summaries with carefully designed data-dependent weights. The resulting estimator is finite-sample instance-optimal and runs in only \(O(\log(n))\) time on sorted samples. --- ### MILER: Semantic Mid-Level Representation for Sim-to-Real Reinforcement Learning in Unstructured Autonomous Driving Authors: Thomas Steinecker, Denis Trescher, Alexander Bienemann, Thorsten Luettel, Mirko Maehlisch Published: 2026-09-17 Categories: cs.RO, cs.LG arXiv: https://arxiv.org/abs/2609.20747v1 Reinforcement learning constitutes a promising approach owing to its potential for superhuman performance and self-learned policies. However, its application to real-world autonomous driving remains scarce, particularly in unstructured environments, because of the challenges associated with sim-to-real transfer for unstructured environments. In this work, we present MILER, an end-to-end policy framework with zero-shot sim-to-real transfer. During offline training, we employ a custom semantic mid-level representation (MLR) simulator and train the policy network using reinforcement learning, with its control outputs applied directly to a bicycle model. During deployment on the real vehicle, camera and LiDAR data are processed by BEVFusion to generate a semantic bird's-eye-view representation consistent with that of the MLR simulator. The actions generated by the policy network are not applied directly to the real vehicle. Instead, we employ a trajectory-alignment strategy that enables zero-shot sim-to-real transfer of both perception and control. We extensively evaluate the proposed framework on a diverse test track comprising numerous challenges, including various obstacles, hairpin curves, velocities of up to 33.6 km/h, and off-road sections. In total, we drove 17.3 km with two different vehicles on a 3.0 km test track without human intervention, thereby demonstrating the effectiveness of our approach. Furthermore, the entire software stack runs on a Jetson AGX Orin. --- ### Video DeltaNet: A Video-Native Hybrid Attention for Livestream Video Generation Authors: Haocheng Xi, Yiming Xie, Hexu Zhao, Yiwen Zhang, Michael Liu, Thomas Creavin, Kurt Keutzer, Xiuyu Li, Zhaoyang Lv, Chenfeng Xu, Haiwen Feng Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20744v2 Video diffusion models repeatedly process long spatiotemporal token sequences during denoising, making attention a major computational bottleneck. Linear attention offers an appealing alternative and has been widely adopted in recent large language models, but directly applying it to video models often fails to preserve the fine-grained interactions required for high-quality generation. We present Video DeltaNet (VDN), which combines local Softmax attention with bidirectional linear memory for long-range video context. Its linear branch introduces Video Delta Attention (VDA), which updates memory once per frame by jointly incorporating its spatial tokens. Separate output projections and learnable gates calibrate the two branches, while a staged teacher-alignment recipe progressively introduces the new pathway into pretrained models. We instantiate VDN on MiniMax H3, applying the hybrid to video-to-video interactions while retaining Softmax for interactions involving text or audio. With eight-step distillation and an optimized SGLang serving stack, VDN-H3 completes DiT denoising for a 14.3-second, 768p video in 6.70 seconds on eight NVIDIA B200 GPUs, corresponding to a 14.5x speedup over the 50-step dense H3 baseline on the same GPU count. GitHub code available at: https://github.com/OpenVDN/vdn-minimax-h3. Weights available at: https://huggingface.co/OpenVDN/vdn-minimax-h3 --- ### Video DeltaNet: A Video-Native Hybrid Attention for Livestream Video Generation Authors: Haocheng Xi, Yiming Xie, Hexu Zhao, Yiwen Zhang, Michael Liu, Thomas Creavin, Kurt Keutzer, Xiuyu Li, Zhaoyang Lv, Chenfeng Xu, Haiwen Feng Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20744v1 Video diffusion models repeatedly process long spatiotemporal token sequences during denoising, making attention a major computational bottleneck. Linear attention offers an appealing alternative and has been widely adopted in recent large language models, but directly applying it to video models often fails to preserve the fine-grained interactions required for high-quality generation. We present Video DeltaNet (VDN), which combines local Softmax attention with bidirectional linear memory for long-range video context. Its linear branch introduces Video Delta Attention (VDA), which updates memory once per frame by jointly incorporating its spatial tokens. Separate output projections and learnable gates calibrate the two branches, while a staged teacher-alignment recipe progressively introduces the new pathway into pretrained models. We instantiate VDN on MiniMax H3, applying the hybrid to video-to-video interactions while retaining Softmax for interactions involving text or audio. With eight-step distillation and an optimized SGLang serving stack, VDN-H3 completes DiT denoising for a 14.3-second, 768p video in 6.70 seconds on eight NVIDIA B200 GPUs, corresponding to a 14.5x speedup over the 50-step dense H3 baseline on the same GPU count. --- ### On-Demand Attention: Language Models Know When to Recall Authors: Haibo Feng, Ruiqi Liang, Dongyang Jin, Hanyang Peng, Shiqi Yu Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20734v2 Reasoning and agentic workloads increasingly demand efficient long-context inference. Yet full-attention decoding reads the growing history at every step, although the benefit of global access varies across prediction positions. We find that, before global attention is computed for the current step, the decoding states available after local computation in frozen pretrained models already contain information predictive of its benefit over local attention. Building on this finding, we introduce On-Demand Attention (ODA), a local-first decoding method: after local computation, a lightweight recall head decides whether to recompute the current step with global attention. ODA trains only the recall head with modest data and compute budgets, leaving pretrained weights unchanged and retaining the complete historical KV cache so that information skipped at one step remains available for later access. Experiments across model scales and families, including hybrid attention backbones, show that ODA recovers most of the performance lost under local attention while substantially reducing the frequency of global attention. Controlled long-context measurements in vLLM further show that GPU-side conditional execution translates fewer global reads into practical decoding speedups over full attention. These findings show that pretrained decoding states can support both token prediction and decisions about accessing distant information, allowing models to allocate global computation as needed during decoding. --- ### On-Demand Attention: Language Models Know When to Recall Authors: Haibo Feng, Ruiqi Liang, Hanyang Peng, Shiqi Yu Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20734v1 Reasoning and agentic workloads increasingly demand efficient long-context inference. Yet full-attention decoding reads the growing history at every step, regardless of its benefit to the next prediction. We show that a pretrained model's decoding states already contain information predictive of this benefit, before the global read. Building on this finding, we introduce On-Demand Attention (ODA), a local-first decoding method that uses a lightweight recall head to selectively invoke global attention as its predicted benefit changes during generation. ODA trains only the recall head, leaving pretrained weights unchanged and the complete historical KV cache available for future recall. We further implement GPU-side conditional execution in vLLM, translating reduced global reads into practical decoding speedups over full attention at long context lengths. Experiments across Qwen and Gemma models, including hybrid-attention backbones, show that selective recall recovers most of the performance lost under local attention while substantially reducing global reads. These findings support long-context inference in which pretrained models guide their own access to the information they retain. --- ### Robust Multi-Task Learning for Principal Component Analysis Authors: Dali Liu, Haolei Weng Published: 2026-09-17 Categories: math.ST, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2609.20733v1 Principal component analysis (PCA) is a fundamental tool for learning low-dimensional structure from high-dimensional data. When data are collected from multiple sources, the underlying task distributions may exhibit unknown degrees of similarity, with some tasks potentially arising from arbitrary distributions. We propose new multi-task PCA procedures that exploit similarity structure across tasks to improve eigenspace estimation while remaining robust to outlier tasks. We establish non-asymptotic convergence rates and show that the proposed procedures attain minimax optimal rates in a range of regimes. One of the procedures builds on the matrix-depth notion of Chen, Gao, and Ren (2018) and can achieve the optimal dependence of the estimation error on the proportion of outlier tasks, addressing a key challenge in robust multi-task learning. Extensive simulations and real-data analyses demonstrate the effectiveness of the proposed methods. --- ### Q&A on Any Spreadsheet Requires Interpreting Its Grid Structure Authors: Zofia Smoleń Published: 2026-09-17 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2609.20732v1 Semantic cell annotation improves chunking interpretability for spreadsheets in LLM-driven RAG systems, aiding answer generation through enriched context rather than improved retrieval accuracy. We propose a novel framework of splitting any spreadsheet into interpretable chunks using cell role annotation. Our framework beats the state of the art, yet it faces a hard ceiling. Spreadsheets are fundamentally two-dimensional unstructured data with continuous relationships and infinite potential cell roles. Because classification models are restricted to finite, pre-defined classes, they cannot perfectly capture this structural nuance, even with human-level annotation. We show that addressing the spreadsheet-to-LLM bottleneck requires moving beyond discrete cell classification. Instead, the field must develop dimensionality-reduction techniques to directly flatten 2D unstructured spreadsheets into 1D unstructured text. Text chunks would be easier for downstream RAG to interpret and generate from. --- ### Deep Noir: Autonomous Steering Discovery via Architectural Chronometry in Transformer Models Authors: Frank E. Bobe, Gregory D. Vetaw, Darshan W. Bryner, Matthew G. Cook, Jose L. Salas-Vernis Published: 2026-09-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.20722v1 Activation steering modifies LLM behavior at inference time, but identifying where and how strongly to steer remains manual. We introduce Deep Noir, a framework that uses Logit Lens convergence and causal head-level attribution to autonomously discover optimal steering parameters. Across three scales (1B x 3, 2-3B x 2, and 7-9B x 4), our engine achieves 16.7 percentage-point improvement on spam at 1B (standard deviation 4.7; 39 runs), with gains increasing to 21 to 42 percentage points at 7-9B across four architectures. On SST-2 sentiment, it achieves a 13.1 percentage-point improvement with zero code changes. Mechanistic grounding enables automated discovery of intervention points that generalize across tasks and architectures. On sentiment, RepE without head masking fails to improve over baseline, while Deep Noir improves all models (p less than 0.01). We further show that steering creates a predictable prompt-injection attack surface whose vulnerability increases monotonically with steering magnitude. This finding is relevant to agent systems deploying steered classifiers. --- ### Don't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL Authors: Juzheng Zhang, Disha Makhija, Manoj Ghuhan Arivazhagan, Vinayshekhar Bannihatti Kumar, Rashmi Gangadharaiah Published: 2026-09-17 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.20715v1 Agent trajectories record what an agent does and what happens next. Yet standard supervised fine-tuning (SFT) applies loss only to agent-authored action tokens, using environment observations as context but not as prediction targets. We ask whether this convention provides the best initialization for subsequent reinforcement learning. We introduce ActObs, which also supervises the observation tokens already present in each trajectory. Although deployed agents never generate observations, learning to predict them encourages the policy to model action consequences without adding data, parameters, sequence tokens, or forward passes. The methods perform similarly after SFT but diverge after GRPO. On Qwen3-4B, GRPO from ActObs achieves higher pass@k at every evaluated sampling budget than its action-only counterpart on Terminal-Bench 2.0. On Qwen3-8B, it trades some pass@1 reliability for higher pass@k (+3.4 pp at pass@16) and solves more distinct tasks. The advantage extends to cross-domain code editing on aider-polyglot (+4.2 pp at pass@1 at 4B), whose tasks are unseen during SFT and RL. ActObs retains more entropy during RL while requiring less policy movement, leaving the final policy closer to its SFT initialization. Our analysis traces this difference to SFT: action and observation gradients rapidly become orthogonal, while action-only training leaves a large residual observation gradient and degrades environment prediction below the base model. Joint supervision prevents this one-sided specialization, preserving consequence prediction and preparing the policy for downstream exploration. --- ### Summarization Bias: The Directional Collapse of Objective Projection into Told-Mode Labels in Large Language Models --- A Conceptual Framework and Registered Test Protocol Authors: Levent Bulut Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20712v1 This paper introduces and operationalizes summarization bias: a proposed systematic tendency of large language models (LLMs) to represent narrative meaning as an abstract summary label rather than as the reconstructable inferential structure that produces it. Within the Bulut Doctrine, narrative effect is theorized along a told-shown axis: in told mode, emotional and informational content is declared explicitly and requires little reader reconstruction; in shown mode, that content is suppressed at the surface and must be reconstructed from physical cues and indirection (Objective Projection). Shown mode is the higher-load condition the doctrine is designed to measure. The claim is that LLMs fail along this axis in a specific direction. Summarization bias is hypothesized to operate in two regimes: (i) a generative regime, in which a model asked to render an emotion through Objective Projection defaults to declaring it instead; and (ii) an evaluative regime, in which a model judging narrative quality rewards told-mode explicitness and under-detects shown-mode suppression. The evaluative regime is the more consequential, since LLMs increasingly serve as judges and reward models, and a directional bias toward told mode would impose a selection pressure degrading prose toward flat declaration. This report does not claim the bias is validated. It defines the construct, situates it against LLM-as-judge biases, rereads a completed independent reliability study as directional evidence consistent with it, and pre-registers a two-regime test with decision rules under which the construct would be abandoned. --- ### Stable Movement for Nondual Lipschitz Convex Optimization: Efficiency and Nearly Optimal Oracle Rates Authors: David Martínez-Rubio, Cristóbal Guzmán Published: 2026-09-17 Categories: math.OC, cs.LG arXiv: https://arxiv.org/abs/2609.20701v1 We study efficient algorithms for realizing the first-order oracle complexity of optimization of $G$-Lipschitz convex functions with respect to the $\ell_{q}$-norm over an $\ell_{p}$-ball of radius $R$, where $1\leq p,q\leq \infty$. For $p<q$, we obtain error $\widetilde{O}_{p,q}(GR/T^{1/p-(1/q-1/2)_{+}})$ after $T$ oracle queries, efficiently realizing the nearly optimal rates of (MBG+26), thereby resolving the nonsmooth end of the COLT 2015 open problem (Guz15b). In particular, the rate is $\widetilde{O}(GR/T)$ for Euclidean Lipschitzness over an $\ell_1$-ball of radius $R$ ($p=1,q=2$). Our solution consists of reducing convex Lipschitz optimization to the chasing nested convex sets problem in sublevel sets of an evolving bundle (LNN95; BBE+20): at each query we either find a point with low function value or we produce a deep cut in the current sublevel of the bundle, that we chase. The dichotomy between stability of selectors and forced movement by deep cuts bounds the number of iterations of the algorithm near optimally. For nested subsets of $R B_{p}^{d}$, we introduce a novel notion of stable center whose movement is bounded by $\widetilde{O}_{p,q}(RT^{1-1/p+(1/q-1/2)_{+}})$ in the $\ell_{q}$-norm after $T$ steps, which we show is nearly optimal in high dimensions. A Monte Carlo average of the proposed selector achieves near-optimal rates with high probability and can be implemented in polynomial time for our optimization algorithm in the real-arithmetic model. --- ### Should This Case Be Adapted? Prediction Fragmentation Controls Test-Time Adaptation Authors: Lili Wang, Jing Li, Xiaowen Sun, Xiangyu Hu, Zhuangzhuang Gu, Jian Liu, Srihari Nelakuditi, Yan Tong Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20700v1 Episodic test-time adaptation resets a frozen segmenter to source weights $M_0$ on each case and adapts for a fixed step count. A fixed horizon conflates a cohort-level question, how far to adapt, with an irreducibly per-case one, whether this case should be adapted at all. Cohort means hide that decision: on cross-vendor cardiac MRI the mean $Δ$Dice from adaptation is statistically indistinguishable from zero while 58.7% of cases are individually made worse. We quantify this harm as harmful accepted area (HA), the harmful fraction of the edited area a controller deploys. Held-out tuning gives a stronger baseline than a fixed horizon, but the budget it selects transfers on neither of the two main medical benchmarks, and no global budget can condition on the case. We show that prediction fragmentation---the disagreement geometry between $M_0$ and the adapted mask $M_k$---predicts HA with no labels or extra backward passes at decision time, comparably on three benchmarks (Spearman $ρ$ 0.50--0.60), at a quarter of gradient-norm's latency. A case-level router built on it cuts HA from 0.228 to 0.139 on a benchmark that took no part in its design, with the design frozen and only cut-points recalibrated there. On the cardiac benchmark the design was selected on, the router cuts HA from 0.129 to 0.013 at matched Dice and 1.10 deployed updates, against the retrospective-best budget found post hoc on evaluation labels, and reduces that 58.7% to 20.0%, an upper bound we quantify. Where the retained cases are not net-helped (as on prostate), the router still cuts HA but concedes accuracy, a boundary we report. Thresholds are fit once on a labeled split disjoint from evaluation; decisions use no labels or gradients. The template ports across architecture and domain (nnU-Net$\to$SegFormer, Cityscapes$\to$ACDC) with coordinate, thresholds and per-bucket actions instantiated per domain. --- ### TetrisCNN for interpretable detection of phases of matter from experimental quantum simulator data Authors: Kacper Cybiński, Björn van Zwol, James Enouen, Guillaume Bornet, Thierry Lahaye, Antoine Browaeys, Antoine Georges, Anna Dawid Published: 2026-09-17 Categories: quant-ph, cond-mat.dis-nn, cs.LG arXiv: https://arxiv.org/abs/2609.20693v1 Detecting phases of matter in general relies on identifying the correct order parameter - a task that remains notoriously difficult for unknown transitions and traditionally is guided by physical intuition and educated guess. Neural networks have recently offered an alternative route by locating phase transitions in known models without any a priori physical knowledge. Yet these approaches remain black boxes and only identify phases without elucidating their properties. Moreover, they often struggle when confronted with realistic, noisy experimental data, which constitute the ultimate testbed for automated methods in physics. Here, we bridge these perspectives by introducing TetrisCNN, a convolutional architecture with parallel branches of differently shaped filters, reminiscent of Tetris blocks, that learns sparse, interpretable latent representations directly in terms of spin correlators. Applied to experimental snapshots of two-dimensional Ising and XY quantum simulators measured in multiple bases, the network not only detects phase transitions and crossovers but also expresses its latent representation and decision boundaries as symbolic formulas built from experimentally measurable spin correlators. This framework opens the way to integrating interpretable neural networks with quantum simulators to uncover and understand new phases of matter. --- ### The First-Order Oracle Complexity of Lipschitz Convex Optimization in Nondual Settings Authors: David Martínez-Rubio, Brian Bullins, Cristóbal Guzmán, Mathieu Molina Published: 2026-09-17 Categories: math.OC, cs.LG arXiv: https://arxiv.org/abs/2609.20687v1 We study first-order black-box convex optimization over an $\ell_p$-ball for objectives Lipschitz in the $\ell_q$-norm, solving in the affirmative the nonsmooth version of the COLT open question (Guz15b) on whether the geometry of a smaller feasible set ($p < q$) can improve convergence rates in convex optimization, and matching prior lower bounds up to logarithmic factors. Our rates include \(\widetilde O(1/T)\) for convex Euclidean-Lipschitz optimization over the $\ell_1$-ball, improving on the $O(1/\sqrt{T})$ classical rate under general assumptions. The key technical device is a new online learning game, where the comparator is evaluated using the maximum of affine losses observed so far. We bound the value of this game above and below in terms of a combinatorial online learning quantity: the sequential fat-shattering dimension, which we characterize for the $\ell_p / \ell_q$ case. Our results generally apply when the feasible set $X$ and the set of possible subgradients $H$ are convex, centrally symmetric, and admit a type of minmax theorem, advancing on a fundamental question by Sridharan (Sri12, Section 10.1.2, Q3). As a geometric consequence of our analysis, of independent interest, we obtain estimates for the expected distance of a convex hull of samples to their mean in several Banach geometries, a version of the celebrated Wendel's theorem (Wen62), but quantitative and for bounded general distributions as opposed to centrally symmetric ones. --- ### HerHealthEval: Evaluating Multilingual and Register-Sensitive Understanding of Women's Health Communication Authors: Hassan Saeed Hassan Albattra, Mazen Mohammed Bahgat, Rahatara Ferdousi, Hana Essam Sayed Ahmed Amrya, Mariam Mousa Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20684v1 Large language models are increasingly used in healthcare communication, yet most evaluations emphasize response quality while assuming that the user's concern has been interpreted correctly. We introduce HerHealthEval, a controlled evaluation framework for multilingual understanding of women's-health communication. For each clinical case, HerHealthEval provides matched versions in English, French, and Modern Standard Arabic using six communicative forms: canonical, clinical, layperson, indirect or hedged, emotionally concerned, and deliberately under-specified. The first five express the same underlying concern and retain the same clinical information, whereas the under-specified form intentionally omits relevant details to test whether the model recognizes that clarification is needed. We evaluate a multilingual instruction model and QLoRA-adapted variants on concern classification, risk calibration, clarification behavior, parse compliance, and cross-form consistency. Results reveal that aggregate accuracy and consistency can conceal safety-relevant failures. A multilingual adaptation model reaches 0.994 under-triage in French and Arabic under language-asymmetric risk supervision. A controlled re-adaptation using source-derived, language-invariant risk labels reduces under-triage to 0.572 and 0.558, respectively. These findings show that robust multilingual healthcare evaluation requires explicit testing of register variation, uncertainty handling, and the provenance and invariance of adaptation labels. --- ### Towards Scaling Marine Perception with Synthetic Data Authors: Haoyu Ma, Onur Bagoren, Anja Sheppard, Elias Fandi, Ashrith Edukulla, Tanner Aslan, Natasha Sieh, Jingyu Song, Katherine A. Skinner Published: 2026-09-17 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2609.20680v1 Scalable machine learning in challenging underwater environments is strongly limited by the lack of labeled real-world training data. This data is often expensive and laborious to gather, making large-scale real-world data challenging to gather and curate. However, simulated data can help close the gap, enabling many learning-based tasks for underwater perception. In this work, we extend OceanSim, an IsaacSim-based underwater perception simulator, with a Synthetic Data Generation (SDG) pipeline for training models to be used in underwater scenarios. The proposed pipeline enables users to generate large, automatically labeled, photorealistic datasets with configurable scene appearance, structure, and sensor settings. We evaluate the pipeline on a real-world sea urchin detection task and study how different forms of synthetic scene variation affect sim-to-real performance. Based on these experiments, we discuss findings on our results, main limitations of the current pipeline and identify future directions for improving underwater rendering fidelity, scene diversity, and the evaluation of sim-to-real generalization. The open-source code can be found at https://github.com/umfieldrobotics/OceanSim. --- ### RISC-V and machine learning: a survey Authors: Shriman Keshri, Apparna Singh, Chinmaya Kumar Palo, Shreya Adya, Subhankar Mishra Published: 2026-09-17 Categories: cs.LG, cs.AR arXiv: https://arxiv.org/abs/2609.20677v1 The intersection of open-source processor architectures and machine learning is driving the demand for customizable, efficient, and accessible hardware. This survey examines the state of the RISC-V ISA in machine learning applications, analyzing current capabilities, challenges, and future directions based on recent research. The analysis covers academic and commercial implementations, software frameworks, and real-world applications. The RISC-V machine learning ecosystem is evaluated, from instruction set extensions and core implementations to compiler optimizations and deployment strategies. Key contributions include a unified taxonomy of RISC-V ML implementations, a comparative analysis of performance and design trade-offs, an evaluation of software toolchain maturity, and the identification of emerging trends in instruction set extensions and specialized accelerators. Findings reveal progress in energy efficiency, specialized instruction development, and framework integration, while highlighting challenges in standardization, verification complexity, and ecosystem fragmentation. The analysis proposes four research directions to address current limitations: specialized neural processing extensions, adaptive and modular processor architectures, security frameworks, and energy-efficient multi-domain architectures. These directions provide a roadmap for advancing RISC-V as a foundational platform for next-generation machine learning systems. --- ### Epidemiological Causal Graph Identification: Challenges, Identifiability and Algorithms Authors: Sambit Mishra, Yingying Wang, Christine K. Johnson, Urbashi Mitra Published: 2026-09-17 Categories: cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2609.20676v1 Causal discovery from observational data is fundamental to statistics and machine learning, yet determining causal direction without interventions necessitates structural assumptions. Existing identifiability research primarily focuses on continuous variables under additive noise models, often neglecting mixed datasets containing ordinal scales, counts, and continuous measurements. This paper investigates causal discovery in Directed Acyclic Graphs (DAGs) where nodes follow either an ordinal distribution (via an ordered logit model) or a regular one-parameter exponential family distribution. We prove that the edge direction between an ordinal and an exponential family node is distributionally identifiable for generic parameter values. Our findings generalize previous Ordinal-Poisson results to the broader exponential family. Computationally, we introduce a score-based exhaustive search and a masked continuous optimization framework using DAGMA for larger graphs. Numerical results validate the theory, recovering edge orientations within a Markov equivalence class that are unidentifiable under classical structural equation models. --- ### FunArt: Decoding Functional Structure and Articulation from Generative 3D Latents Authors: Dennis Rotondi, Abdelrhman Werby, Kai O. Arras Published: 2026-09-17 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2609.20673v1 To operate effectively in human environments, robots must identify articulated objects, segment their movable and interactive parts, and estimate their kinematic models. Existing articulated scene representations typically recover kinematics from observed interactions, while methods operating on static scans often decouple articulation from functional interactive elements. We present FunArt, a framework that constructs articulation-aware functional 3D scene graphs from posed RGB-D observations captured in a single static configuration. FunArt reconstructs object instances, converts their fused geometry directly into the O-Voxel representation of TRELLIS.2, and exploits its frozen, sparse-compression VAE as a structural prior. A lightweight query-based decoder combines compact object-level latents with dense, surface-aligned features to jointly segment movable parts and functional interactive elements while estimating motion type, axis, origin, and range. On the Articulate3D dataset, FunArt achieves state-of-the-art performance across movable-part segmentation, articulation estimation, and functional-element segmentation, both with and without ground-truth object input. In the end-to-end setting, it outperforms the strongest baselines by 1.5 AP_{50} points for movable parts, 2.8 AP_{50} points under joint origin-and-axis constraints, and 6.7 AP_{50} points for functional elements. These results demonstrate that generative 3D latents encode actionable structural cues that can initialize robotic perception and planning before physical interaction. --- ### Learning Foresight without Explicit Trajectories for 3D Diffusion Policies Authors: Zhongbo Zhang, Zaibin Zhang, Yifan Wang, Changbo Yan, Lijun Wang, Huchuan Lu Published: 2026-09-17 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2609.20669v1 3D diffusion policies are strong at generating geometrically grounded actions from current observations, but successful manipulation requires not only knowing what motion is feasible now, but also anticipating where the interaction is heading. Existing policies largely leave such foresight to emerge implicitly from action learning. We introduce Movement Trend Guidance, a simple but effective way to provide this foresight without introducing an explicit plan. From a short observation history, the policy learns a compact latent representation of interaction evolution. During training, sparse future gripper states supervise this representation; at inference, only the latent is retained as future-oriented conditioning alongside the current observation. The latent provides global conditioning for action generation, while an additional gated FiLM branch is used only at the UNet bottleneck. Despite adding only 3.52% more parameters to DP3, our method preserves the original dense-action and receding-horizon formulation and consistently improves upon DP3 across RoboTwin2.0, LIBERO-40, and DexArt. It reaches 62.8% vs. 56.1% in 50-task RoboTwin2.0 mixed training, 71.93% vs. 37.08% on LIBERO-40, and 72.0% vs. 49.0% on five real-robot tasks. These results show that a diffusion policy can benefit substantially from knowing where an interaction is heading, without being told exactly where to move. --- ### Earth Surface Immune System for Rapid Monitoring of Unknown Anomalies Authors: Jingtao Li, Qian Zhu, Xinyu Wang, Deren Li, Liangpei Zhang, Yanfei Zhong Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20662v1 Earth surface anomalies, driven by escalating climate change, and expanding human activities, are increasing in both frequency and diversity, yet their limited historical data and unpredictability make them fundamentally different from conventional remote sensing targets. Existing methods address specific anomaly categories or stop at localization, leaving a gap between detection and actionable information. Here we present ESIA, an Earth Surface Immune System whose architecture is constrained by three principles from the biological immune system, refined over millions of years against equally diverse and uncertain threats. A non-specific innate immune stage treats anomalies as unobserved changes in time-series satellite imagery, generating binary localization maps at 14.51 km2/s without assuming any anomaly category, surpassing the strongest general baseline by 37% in F1. A specific adaptive immune stage applies negative selection to filter text prompts and matches surviving prompts with localized image patches through a multi-modal foundation model, enabling open-vocabulary recognition of unknown anomaly attributes including category, affected area, and damage severity, with recognition F1 exceeding 80%. A mutation mechanism tunes minimal embeddings at test time, adapting to each scene in 3.26s using a single reference image pair. We validate ESIA on a global-scale dataset covering 19,801.60 km2 across six anomaly categories, comparing against 22 models, and further apply it to quantify degraded farmland in the Dnipro Delta following the Kakhovka Dam collapse and assess burn severity from 2025 Palisades Fire in Los Angeles. This unprecedented flexibility in handling unknown anomalies opens new avenues for real-time disaster response and environmental surveillance. --- ### HIL-UMI: Bringing Human-in-the-Loop Post-Training of Vision-Language-Action Models to Universal Manipulation Interface Authors: Zimu Han, Yiming Zeng, Jiyao Zhang, Zihao Zhao, Yuanfei Wang, Yixiang Jin, Shiqi Li, Shuangben Chen, Wei Huang, Ruodai Li, Hui Shen, Hao Dong Published: 2026-09-17 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.20659v1 Large-scale vision-language-action (VLA) models provide powerful priors for robot manipulation, yet adapting them to a specific deployment remains challenging. Supervised fine-tuning (SFT) on task-specific demonstrations provides a step toward deployment, but faces two persistent limitations: static data provide limited coverage of out-of-distribution states, and standard imitation objectives do not distinguish progressing behavior from less useful data. Interactive post-training can address these limitations, but typically requires repeated policy execution and human intervention on a physical robot. We introduce HIL-UMI, a policy-guided Universal Manipulation Interface (UMI) framework for robot-free human-in-the-loop VLA post-training. During handheld UMI demonstrations, HIL-UMI queries the current policy on the same observation stream without executing its predictions. The Energy Score compares the human action trajectory with policy inference and triggers collection when their discrepancy indicates an out-of-distribution region. In a separate feedback loop, low online advantage predictions identify essential segments for refining a progress-based advantage estimator. The updated estimator then guides advantage-conditioned behavioral cloning using a balanced mixture of base demonstrations and new policy data. This design preserves the iterative and policy-aware nature of human-in-the-loop learning while decoupling data collection from robot deployment. Experiments on four real-world tasks spanning long-horizon and precise manipulation show that HIL-UMI achieves consistent improvement over SFT and benefits from both targeted collection and advantage refinement. Moreover, HIL-UMI outperforms HG-DAgger on Clean Up Table with lower per-frame collection time, suggesting a scalable path for VLA post-training across operators and locations. --- ### Ownership in AI-Assisted Everyday Tasks Authors: Megan Wei, Melanie Subbiah, Audrey Lee, Annya Dahmani, Dave Edwards, Helen Edwards, Ellie Pavlick Published: 2026-09-17 Categories: cs.AI, cs.HC arXiv: https://arxiv.org/abs/2609.20658v1 When does work done with AI still feel like ours? As AI becomes woven into everyday tasks, we must examine what happens to our sense of ownership and contribution when a machine shares in producing what we make. We report an exploratory qualitative survey in which participants were asked to describe two recent, self-selected tasks completed with AI: one that felt like their own and one that did not. We find that felt ownership depends on the process of collaboration: people disown work when they merely approve AI's suggestions, but retain ownership when they lead, iterate, or rewrite. Ownership can also extend to settings where people own the vision for a project but not the execution; respondents reported high ownership on tasks they could not have completed without AI. Loss of personal voice and a lack of comprehension of the output both erode ownership. Finally, willingness to disclose AI use is often decoupled from actual pride or ownership, and instead shaped by community norms and fear of credit erasure. We propose several research directions as a result of these findings to promote AI development that supports people's sense of authorship over their own lives. --- ### Multi-center Medical Data Mining with FL-Net - A One-stop Shop for Federated Learning Authors: Simon Süwer, Julian Klemm, Elisa Acitelli, Mathieu Almeida, Lucia Altucci, Zsolt Bagyura, Michelangela Barbieri, Zsolt-Zoltán Bedő, Rosaria Benedetti, Béla Bihari, Csongor Csalóka, Lucia Dicunta et al. Published: 2026-09-17 Categories: cs.LG, cs.CR, cs.DC arXiv: https://arxiv.org/abs/2609.20650v1 Federated learning enables collaborative training without sharing patient-level data, but most studies remain simulations. Based on five requirements derived from the literature, we analyzed 14 FL frameworks and found that none fully satisfied these requirements. We present FL-Net, a novel federated clinical research framework to fulfill all requirements. It integrates modular data harmonization, data discovery, disclosure control, securely built versioned FL-Net-Tools and containerized federated workflow execution into a persistent network. It enables the re-use of harmonized data and workflows across studies. FL-Net's end-to-end capabilities were evaluated through harmonization, cross-study patient discovery across MIMIC and US-130, and reproducible, audited federated workflows with up to 50 concurrent clients. FL-Net is being developed within the dAIbetes and Microb-AI-ome EU projects and will cover over 800,000 patients across 10 hospitals in 9 countries covering longitudinal and single point in time data, FL-Net provides a practical foundation for interoperable, reproducible, and privacy-preserving multicenter clinical research. --- ### DexTouch-WM: Learning Action-Conditioned Tactile World Models from Human Touch for Dexterous Robot Manipulation Authors: Yan Qin, Yue Chen, Wenwei Lin, Shujia Liu, Chuqiao Lyu, Kailun Su, Weiyang Jin, Chenze Yu, Ping Luo, Wenbo Ding, Tianxing Chen, Renjing Xu Published: 2026-09-17 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2609.20649v2 Learning predictive models of contact-rich dexterous manipulation requires dense tactile interaction, but such data are costly to scale on real robots and remain tied to embodiment-specific sensors. We introduce DexTouch-WM, an action-conditioned world model that learns from scalable human touch to jointly predict future RGB observations and bilateral tactile dynamics. Our insight is that human and robot manipulation share transferable contact dynamics when their tactile observations and action spaces are made compatible. We deploy flexible piezoresistive arrays with a shared sensing layout on both human and dexterous robot hands, and retarget human motion into the robot action space so that human interaction can supervise the same dynamics model used for real-robot prediction. DexTouch-WM couples a pretrained video expert with a lightweight tactile expert using anatomy-aware tactile tokens and aligned action conditioning. In human-to-robot scaling experiments, we keep five hours of real-robot supervision fixed while increasing human interaction from 0 to 100 hours, and observe substantial improvements in held-out robot-domain visual, geometric, and contact prediction despite disjoint human and robot task sets. Beyond prediction, we evaluate the world models as surrogate environments for policy evaluation and as generators of synthetic trajectories for real-robot policy learning, showing that scalable human interaction provides a complementary data axis for learning dexterous robot world models. --- ### DexTouch-WM: Learning Action-Conditioned Tactile World Models from Human Touch for Dexterous Robot Manipulation Authors: Yan Qin, Yue Chen, Wenwei Lin, Shujia Liu, Chuqiao Lyu, Kailun Su, Chenze Yu, Ping Luo, Wenbo Ding, Tianxing Chen, Renjing Xu Published: 2026-09-17 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2609.20649v1 Learning predictive models of contact-rich dexterous manipulation requires dense tactile interaction, but such data are costly to scale on real robots and remain tied to embodiment-specific sensors. We introduce DexTouch-WM, an action-conditioned world model that learns from scalable human touch to jointly predict future RGB observations and bilateral tactile dynamics. Our insight is that human and robot manipulation share transferable contact dynamics when their tactile observations and action spaces are made compatible. We deploy flexible piezoresistive arrays with a shared sensing layout on both human and dexterous robot hands, and retarget human motion into the robot action space so that human interaction can supervise the same dynamics model used for real-robot prediction. DexTouch-WM couples a pretrained video expert with a lightweight tactile expert using anatomy-aware tactile tokens and aligned action conditioning. In human-to-robot scaling experiments, we keep five hours of real-robot supervision fixed while increasing human interaction from 0 to 100 hours, and observe substantial improvements in held-out robot-domain visual, geometric, and contact prediction despite disjoint human and robot task sets. Beyond prediction, we evaluate the world models as surrogate environments for policy evaluation and as generators of synthetic trajectories for real-robot policy learning, showing that scalable human interaction provides a complementary data axis for learning dexterous robot world models. --- ### Beyond PINNs: A Unified Gauss--Newton and Petrov--Galerkin Framework for Neural and Hybrid PDE Solvers Authors: Nilo Schwencke, Roland Maier Published: 2026-09-17 Categories: math.NA, cs.LG arXiv: https://arxiv.org/abs/2609.20641v1 Physics-informed neural networks and finite element methods provide two different paradigms for the numerical approximation of partial differential equations: the former are commonly trained by minimizing pointwise strong residuals, whereas the latter are naturally built from weak variational formulations and the finite-dimensional systems obtained after discretization. In this work, we introduce a common framework based on the discretization of functional Gauss--Newton problems by finite families of linear measurements. We show that, through an appropriate duality pairing, the linear measurements can be represented by test functions. The resulting Gauss--Newton system is then precisely a Petrov--Galerkin discretization of the linearized functional problem. This perspective recovers pointwise collocation and natural-gradient constructions as particular cases, while making the choice of test functions an explicit algorithmic design choice. We specialize this framework to elliptic problems, where it naturally leads to weak residual formulations and to a hybrid finite element--neural construction acting on complementary approximation spaces. Numerical experiments support the proposed framework and demonstrate the effectiveness of weak Gauss--Newton formulations and hybrid finite element--neural approximations. --- ### PROVIA: Procedure State Tracking for Online Mistake Detection in Egocentric Videos Authors: Di Wen, Kailun Yang, Jimmy Weissert, Luc Maria Scherrer, Cedric Zöllner, Ruiping Liu, Yufan Chen, Jiale Wei, Junwei Zheng, Kunyu Peng Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20638v1 An assistant watching egocentric video should notice a mistake from past frames alone, before the next step begins, and keep working once the person recovers. A mistake changes the state of the work, so every later step has to be read against what was done rather than against the plan. The first-mistake protocol that current online methods report on cuts each recording at its first mistake, so a fixed-time rule that never looks at the video is right on every case. We evaluate on complete trials, where mistakes and recoveries arise naturally, under a validation false-alarm budget and against controls that use timing alone. PROVIA keeps two records apart: a factual state, a learned summary of the steps each actor performed, mistakes included, and the accepted progress, an exact posterior over the state of an automaton induced from correct demonstrations by Bayesian state merging and over the execution status of each actor. Procedure-state transitions occur only in the correct-status branch; the mistake and correction branches retain the source state. A sequential test turns the per-frame mistake probability into alarms. With one filter and one optimization rule, PROVIA ranks mistakes best among the evaluated controlled baselines on CaptainCook4D, IndustReal, HoloAssist and IMPACT-ego. At a validation budget of 0.1 false alarms per minute it recalls .154 against .128 on CaptainCook4D and .034 against .015 on HoloAssist, where it leads at every budget. The pipeline runs at 58-70 frames per second. The source code is available at https://github.com/Kratos-Wen/PROVIA. --- ### PAA: The Probabilistic Allen Algebra: A Generative and Complete Probabilistic Extension of Allen's Interval Relations Authors: Julian Eggert Published: 2026-09-17 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.20634v1 Allen's interval algebra is a qualitative calculus for temporal relations, but its thirteen base relations are crisp predicates over exact interval boundaries. This is inadequate for temporal information from language, perception, databases, or uncertain histories, where times, durations, and boundaries are uncertain and expressions such as "just before" or "roughly during" have graded meaning. We develop the probabilistic Allen algebra (PAA): a generative and complete extension in which relation probabilities are derived from distributions over interval boundaries rather than assigned as scores. Time points are Gaussian; intervals have Gaussian midpoints and truncated-Gaussian durations. Every relation is a boundary-ordering predicate in one common probability space: point-point relations reduce to error functions, and point-interval and interval-interval relations to multivariate Gaussian orthant probabilities induced by linear inequalities. Contact relations (meets, starts, finishes, equals) receive positive measure through a tolerance band, and under a single tolerance the thirteen relations form a true partition that recovers crisp Allen as the tolerance vanishes. The construction derives Allen's taxonomy rather than positing it: coarse predicates such as precedence, overlap, and containment are unions of leaves whose probabilities are leaf sums, and this hierarchy is preserved as intervals collapse to points and thirteen relations reduce to five and then three. Each relation further decomposes into correlation-aware temporal primitives in the spirit of CIDOC CRM. The algebra is scale-invariant and separates graded expressions such as "shortly before" from contact relations. All results are Monte-Carlo validated and shipped as an open, tested Python package. --- ### Refinement Is Inherently Editable: Training-Free Prompt-to-Prompt Image Editing with Generative Refinement Network Authors: Yulong Chen, Ziqian Zhang, Haoyu Zhang, Ao He, Yaxing Wang, Senmao Li, Kai Wang Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20633v4 Text-guided image editing must introduce the requested changes while preserving unrelated source content. In training-free editing, diffusion editors often use spatial controls whose inaccuracies can leave edits incomplete or alter unrelated regions. Causal autoregressive editors face a further constraint: their fixed decoding order limits revision of earlier decisions. As the first to explore training-free image editing with Generative Refinement Networks (GRN), we observe that its refinement process is inherently suitable for editing and offers a promising way to address these limitations. Motivated by this observation, we introduce RefineEdit, a training-free prompt-to-prompt image editing framework built on the GRN. Our key idea is to couple edit localization with content generation through the global refinement of binary image codes, allowing editing evidence to be revised as the image evolves. More specifically, RefineEdit combines bit routing with two stabilization mechanisms: adaptive spatial freezing and finite bit locking. Bit routing starts from an intermediate source state and uses signed probability differences between the two branches to identify editable positions and bits. It directs selected bits toward editing refinement while anchoring the rest to the evolving source trajectory. Adaptive spatial freezing limits unnecessary expansion of the editing region, while finite bit locking maintains recent bit activations to support continued editing. The overall framework requires no additional training, external masks, or attention control. Across nine editing categories of PIE-Bench, RefineEdit achieves the best background-preservation scores in PSNR, LPIPS, MSE, and SSIM, together with the highest whole-image and edited-region CLIP scores among the evaluated methods. Code is available at https://github.com/mura1n/RefineEdit. ---