# Research Papers — 2026-10 · bundle page 92 of 156

> Full-text markdown of papers #4551–4600 of 7799 published in 2026-10, newest first. AI-readable mirrors of the full 7799-paper 2026-10 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-10/page/92
[← Previous bundle](https://aipod.dev/papers/archive/2026-10/page/91.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-10/page/93.md)

### Large Language Model-Assisted Preparation of Transportation Management Plans: A Case Study with WisDOT WisTMP System

Authors: Zihao Sheng, Pei Li, Zilin Huang, Yen-Jung Chen, Yuhao Luo, Zhengyang Wan, Steven T. Parker, David A. Noyce, Sikai Chen
Published: 2026-10-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.10650v1

Work zones are critical yet hazardous components of transportation infrastructure, requiring carefully designed Transportation Management Plans (TMPs) to ensure safety and mobility. However, TMP preparation remains labor-intensive and heavily dependent on practitioner expertise. This paper proposes a Large Language Model (LLM)-assisted framework to automate TMP content generation, leveraging the WisDOT WisTMP system as the application context. The framework fine-tunes multiple open-source LLMs across different model scales and deploys them locally to ensure data security. To support model training, we construct a domain-specific dataset from historical WisTMP documents by converting PDF files into structured question-answer pairs in JSON format. Experimental results show that fine-tuning significantly improves performance across standard text generation metrics. Further section-wise and strategy-level analyses reveal that, while LLMs achieve strong overall performance, they tend to over-generate strategies and struggle to produce project-specific justifications and accurate cost estimates. In addition, scaling from 7B/8B to 14B yields limited gains. These findings demonstrate the potential of LLMs to improve TMP preparation efficiency while highlighting remaining challenges in LLM-assisted TMP development. The source code and demo videos will be publicly available at https://zihaosheng.github.io/TMP-LLM/.

---

### Leakage-Controlled Multimodal Learning for Diagnosis and Progression Prediction in Alzheimer's Disease Research

Authors: Akeem Temitope Otapo, Ghazaleh Khodabandelou, Zuheng Ming, Alice Othmani
Published: 2026-10-07
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.10648v1

Alzheimer's disease prediction involves irregular visits, heterogeneous measurements and incomplete modalities. This study presents a multimodal multitask framework combining an adapted SFCN MRI encoder, four causal clinical Transformers, shared fusion and task-specific ODE-GRU dynamics. Fine-tuning and LoRA adapt the final two MRI blocks. Task-DRO balances task losses, while Group-CVaR targets cohort and comorbidity strata. Branch-specific input controls, subject-grouped partitions and empirical causality checks support longitudinal evaluation. Across 2,649 subjects and 17,317 visits from ADNI, OASIS-2 and MIRIAD, internal validation yields diagnosis, stage-1 progression and first-stage-1-visit progression AUROCs of 0.935 +/- 0.002, 0.884 +/- 0.003 and 0.870 +/- 0.005, respectively (mean +/- SD across three seeds). Corresponding hybrid AUROCs are 0.951, 0.909 and 0.896. Next-visit MMSE mean absolute error (MAE) is 1.61 points; worst-stratum diagnosis AUROC is 0.827 +/- 0.008. Sampled ADNI explanations identify task-specific input dependence. OASIS-3 external validation yields network and hybrid diagnosis AUROCs of 0.763 and 0.767, hybrid next-visit progression AUROC of 0.764, diagnosis calibration error decreasing from 0.197 to 0.052, and next-visit MMSE MAE of 0.86. Seed-42 paired ablations of six components yield pooled diagnosis and progression AUROC differences between -0.004 and +0.004; removing clinical encoder inputs lowers diagnosis AUROC by 0.272. The framework integrates longitudinal prediction, missing-modality handling, auxiliary comorbidity modelling and subgroup evaluation within a common pipeline.

---

### Masked Generative Motion Planning with Geometry-Guided Token Search

Authors: Lipeng Zhuang, Yingdong Ru, Shiyu Fan, Edmond S. L. Ho, Gerardo Aragon Camarasa, Paul Henderson
Published: 2026-10-07
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2610.10646v1

Generative motion planners typically use learned trajectory priors for initial generation, while leaving test-time repair to local continuous refinement. We introduce Masked Generative Motion Planning (MGMP), which extends the learned prior from efficient parallel generation to structural repair. A masked generative transformer generates discrete trajectory candidates in parallel, and Geometry-Guided Token Search (GGTS) uses scene geometry to target where to edit and which prior-supported alternatives to evaluate. This turns refinement into an efficient search over discrete motion alternatives, enabling route-level restructuring beyond local trajectory deformation. MGMP achieves 96% success on Ring Maze and 82% repair success on Controlled Route Invalidation on Kuka, exceeding the strongest external baselines by 23 and 25 percentage points, respectively. It further generalizes to unseen layouts, additional obstacles, unseen geometries, single- and dual-arm planning, and real-world Baxter tasks.

---

### From Log-Odds to Shapley Values: An Explanatory Geometry for the Weighted Naive Bayes Classifier

Authors: Vincent Lemaire, Fabrice Clérot
Published: 2026-10-07
Categories: stat.ML, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.10642v1

This paper studies the construction of an explanatory space for a weighted naive Bayes classifier from the supervised representation induced by the model. We start from the classical supervised distance based on conditional log-likelihoods and introduce a discriminative reformulation based on log-odds, which is more directly related to the classification decision. We then show that this representation induces a distance that exactly coincides with the $\ell_1$ distance between vectors of analytical Shapley values, thereby providing a formal explanatory interpretation of the geometry induced by the model. Finally, we empirically compare several supervised distances derived from these representations using a $k$-nearest neighbors classifier. This work highlights a close link between supervised distance, local explanation, and predictive behavior, from a primarily methodological perspective.

---

### Coverage-Aware Reasoning with Medical Tokens for Diagnosis Prediction

Authors: Kaisong Zhang, Haotian Fang, Junmeng Zhou, Hang Lv, Yulan Pan, Yanchao Tan
Published: 2026-10-07
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.10641v1

Large language models (LLMs) offer promising potential for next-visit diagnosis prediction, owing to their ability to integrate longitudinal clinical evidence and reason over it in natural language. However, reinforcement learning for LLM reasoning commonly rewards each trajectory according to the correctness of its final answer. In next-visit diagnosis prediction, multiple diagnoses can be simultaneously valid, but independently rewarding one diagnosis per trajectory does not distinguish repeated hits from coverage of different diagnoses. The policy can therefore concentrate on a few correct diagnoses, leaving others uncovered. Meanwhile, LLM tokenizers can split ICD codes into several generic tokens with limited clinical meaning, requiring multiple decoding steps to predict each diagnosis and hindering reasoning over a large disease vocabulary. To address both challenges, we propose CARing, a framework that represents diagnoses with compositional Semantic IDs (SIDs) and optimizes reasoning trajectories for multi-label coverage. Concretely, we first encode ontology-enriched disease semantics into compact SIDs through residual quantization, and ground the resulting SID tokens in natural language and longitudinal EHR contexts through multi-task alignment and reasoning-enriched training to unlock transferable LLM reasoning. CARing further improves unordered multi-label prediction through a coverage reward for reinforcement learning and multi-positive supervision. At inference time, the model supports both efficient direct constrained decoding and multi-chain reasoning with rank fusion. On MIMIC-III and MIMIC-IV, CARing exceeds all EHR-trained baselines in weighted F1 and attains the highest top-k recall at every reported cutoff, including R@30 of 46.04% and 46.52% in reasoning mode. Our codes and logs are available at https://github.com/zmlxzyh/CARing-Codes-Logs.

---

### How Many Directions Must a Truncated Diffusion Sampler Retain? Matching Bounds Under Power-Law Spectra

Authors: Radmehr Karimian, Ali Mohades, Johannes Lederer
Published: 2026-10-07
Categories: math.ST, cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2610.10640v1

Diffusion samplers can reduce computation by generating selected spectral coordinates and filling the remaining directions with noise. How many directions must they retain? We study this question for data with power-law covariance spectra. For Gaussian data compared to a smoothed target, we prove matching bounds on the required number of retained directions, provided that the ambient dimension is sufficiently large. The truncation error depends on the combined Wiener gains of the omitted directions, regardless of the accuracy of the sampler on the retained coordinates. Keeping only directions whose signal exceeds the output noise level can therefore leave a non-vanishing error: many individually weak directions remain significant in aggregate. Combining this characterization with a diffusion convergence bound yields sufficient sampling-step complexity under exact scores. The upper bounds also extend to estimated principal components and, componentwise, to Gaussian mixtures. The practical prescription is to select the retained subspace using an aggregate spectral-tail error budget, then to choose the diffusion noise level accordingly.

---

### Visible Reasoning Is Not a Universal Optimizer: Persona- and Thinking-Dependent Effects in Analytics Code Generation

Authors: Bhawani Shankar Leelar, Pawan Chorasiya, Davin Hill, Robert E. Tillman, Tamer Soliman
Published: 2026-10-07
Categories: cs.LG, cs.AI, cs.SE
arXiv: https://arxiv.org/abs/2610.10639v1

Visible Chain-of-Thought (CoT) is often treated as a broadly useful reasoning instruction, yet analytics code generation combines natural-language ambiguity, schema grounding, target-language constraints, and model-specific inference behavior. Because the same analytics request can be expressed in two distinct target languages-SQL and Python (pandas)-this setting provides a natural test of a common but under-examined assumption: that visible reasoning is more effective when its representation matches the requested target, as in "think in SQL" or "think in Python." Together with generic instructions such as "think step-by-step," such recommendations remain insufficiently evaluated under controlled, execution-based comparisons. We study a query matched SQL-pandas benchmark that crosses persona phrasing, target language, visible-CoT format, control prefixes, direct generation, and internal-reasoning configurations. The results do not support either a universal accuracy advantage from visible CoT or a consistent benefit from matching the reasoning representation to the target language. Instead, the effects depend on the persona, target, model configuration, and internal-reasoning setting. The control ablations further distinguish effects of reasoning content from those of prompt format. These findings indicate that reasoning strategies should be selected jointly for the model, persona, target, and internal-reasoning configuration rather than adopted as universal defaults. More broadly, the study provides a controlled framework for identifying when visible reasoning improves executable generation, when it primarily perturbs model behavior, and when the internal-reasoning configuration is the more consequential factor.

---

### D-SLR: The Disjoint Row-Sparse plus Low-Rank Decomposition

Authors: Vincent Szolnoky
Published: 2026-10-07
Categories: cs.LG, stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2610.10636v1

Compressing a matrix for reconstruction still defaults to the truncated SVD, approximating the data with a single low-rank structure. It is common to reduce the residual further by adding an overlapping row-sparse component, but methods that solve this joint problem often require iterative solvers and tuning of regularization parameters. We propose the Disjoint Row-Sparse plus Low-Rank (D-SLR) decomposition, a closed-form drop-in for the truncated SVD that improves or exactly matches it. D-SLR restricts rows to either being stored verbatim or approximated by the low-rank fit, never both. Under squared error this restriction costs nothing: the joint optimum is attainable disjointly with fewer parameters at every non-trivial rank and stored row count (shape). With zero stored rows D-SLR reduces to the truncated SVD, so it never does worse at equal cost. The algorithm scores the entire error-versus-parameters tradeoff, and the solution is chosen afterwards by a supplied error target or parameter count, or by a selection rule. The grid and solution together cost three SVDs, with no tuning or regularization. We derive an assumption-free, a-posteriori lower bound on the error at every shape, giving each solution a computable certificate on the potential gain of any other choice of rank and stored rows. Experiments on synthetic and real data (LLM embedding tables, network traffic, hyperspectral images) confirm the gains and quantify the certificate.

---

### Speaking the Navigator's Language: Trajectory-Grounded Instruction Translation for Frozen Aerial VLN Agents

Authors: Xi Chen, Zhe Liu, Xiaogang Xu, Jiafei Xu, Chunyi Zhou, Yuan Su, Rui Zeng, Tianyu Du, Kelu Yao, Chao Li, Shouling Ji
Published: 2026-10-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.10635v1

Aerial vision-and-language navigation (VLN) agents are typically trained on detail-rich, trajectory-aligned commands, whereas users issue short, intent-driven instructions; on a frozen OpenFly navigator, this \emph{instruction gap} drops success rate (SR) from $31.03\%$ to $11.33\%$. To scale translator training, we prompt a language model with human-written style examples to convert original commands into paired, intent-centered Weak commands, which yield $15.27\%$ SR. We introduce the \textbf{Trajectory-Grounded Instruction Translator (TGIT)}, a front-end that keeps the navigator frozen and translates Weak inputs into agent-executable commands by learning from its trajectory outcomes. The resulting Weak-trained translator raises Weak-input SR to $37.93\%$ and transfers zero-shot to real human instructions ($11.33\%{\rightarrow}32.51\%$); it also improves held-out OpenFly ($4.95\%{\rightarrow}20.79\%$) and yields recovery on CityNav and AirVLN.

---

### Has LLM Screening Performance Stalled in Software Engineering Systematic Reviews?

Authors: Aleksi Huotala, Miikka Kuutila, Mika Mäntylä
Published: 2026-10-07
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2610.10633v1

Screening in systematic reviews (SRs) is manual and time-consuming. Prior work has explored large language models (LLMs) for automating this step, but LLMs are evolving rapidly, so earlier performance claims may no longer accurately reflect their screening performance. We used an existing software engineering SR screening benchmark (SESR-Eval) as our data. We also power-sampled a new, smaller dataset (SESR-Eval-Mini) that allows evaluation at lower costs. Using this data, we evaluated eight new LLMs for screening performance. Additionally, we tested different prompts, analyzed LLM agreement in screening decisions and criteria, and examined the effect of refining the inclusion and exclusion criteria on screening performance. The eight new LLMs performed marginally better than the seven old ones: avg. MCC across secondary studies rose from 0.347 to 0.365. Differences between secondary studies are still bigger than between LLMs. Computing the overall screening decision from criterion-level decisions degraded screening performance only slightly. LLMs generally agree with each other in their corresponding screening decisions (mean Gwet's AC1 = 0.830), though certain inclusion and exclusion criteria showed larger disagreement than others. Refining the inclusion and exclusion criteria slightly improved recall and made decisions easier for some LLMs, but overall impacts of criteria refinement were modest. LLMs are not yet ready to replace humans in paper screening and the advantages new, more costly models bring, appear to be very limited. Agent-based approaches, prompt engineering, and further criteria refinement are three potential future research avenues.

---

### Phase-HDC: Replacing Optimizer History with Gradient Thresholds in Discrete Phase Learning

Authors: Ahmed Nebli
Published: 2026-10-07
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.10630v1

Training a compact model often needs far more memory than storing it, because the optimizer keeps its own records of past gradients. For a hyperdimensional classifier whose learned parameters are low-bit angles, which we call a \emph{phase memory}, these records take several times more memory than the model itself. We ask whether such a model can be trained while storing nothing but the model. The proposed method, Phase-HDC, turns each stored angle by at most one step per update, against the sign of its current gradient, and only when that gradient is large enough. We show that this simple rule is the exact solution of a first-order loss model in which every changed parameter pays a fixed cost. When everything except the update rule is held fixed, Phase-HDC matches the accuracy of Adam with 6-bit moments while storing three times less. Across eleven image, tabular, and text datasets, it stores 16--23$\times$ less than standard float32 Adam and 4--6$\times$ less than 8-bit Adam. The price is an average loss of about five accuracy points against float32 Adam, while Phase-HDC is more accurate than 8-bit Adam on six of the eleven datasets, including byte-level text prediction, where 8-bit Adam collapses. Instrumented training runs explain these outcomes. Once parameters must sit on a discrete grid, Adam's moments mainly decide whether a parameter moves at all, a decision that a threshold on the current gradient can make without memory, and coarse quantization of the moments breaks this decision for inputs that the data rarely contain. The storage savings are logical state rather than measured hardware memory.

---

### The Harness as the Only Mutable Surface: Compliance-Bounded Self-Evolution of LLM Agents in Credit Pipelines, with a Measured Admission Gate

Authors: Ravil Akhtyamov
Published: 2026-10-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.10629v1

Self-improving LLM agents can adapt a credit pipeline to a changed rule, but an agent that rewrites itself destroys the artefact a supervisor reviews: a named change, a recorded test, an approval. We argue that self-evolution is reviewable only if it is confined to the runtime harness (instruction text, tool-call logic and primitive composition) while model weights stay fixed, so that every adaptation is a diff with a cause and a test attached. We give a dual-loop engine built on that bound, with one admission gate that writes a hash-chained record before deployment, and we measure the gate in simulation, with a simulated agent and a seeded-search proposer rather than language models. Across three families of supervisory re-interpretation at three severities, 10 seeds each, the gated loop admitted 144 of 7,449 candidate changes, none of which worsened error on held-out history, and restored the false-positive rate to the oracle level without raising missed flags in every low- and mid-severity cell. With the gate replaced by the check an unbounded system applies (fewer errors visible in recent traces), the same loops admitted 309 harmful changes and left missed flags above 10% in 49 of 90 runs: false positives fell because the screen was loosened. Evaluated on pre-shift labels, the gate rejected every candidate, so a re-interpretation must be encoded as a rule that relabels history. Parametric and scope shifts were repaired locally, a structural one only by primitive replacement; at the highest structural severity the gate's fixed tolerance blocked the correct replacement in half the seeds. We map the mechanisms to the EU AI Act's provisions for high-risk credit scoring and note that the April 2026 US model-risk guidance excludes agentic AI from its scope.

---

### Agent4RE: A Self-Refining Multi-agent Framework for End-to-End Software Requirements Engineering and Benchmarking

Authors: Yongjian Tang, Linhan Li, Thomas Runkler
Published: 2026-10-07
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2610.10628v1

Existing LLM-based approaches for software Requirements Engineering (RE) typically rely on basic prompting strategies or rudimentary agent collaboration, under-utilizing the full potential of multi-agent systems. Meanwhile, available datasets focus on isolated subtasks, such as requirements extraction, classification, and completeness detection, leaving the absence of an end-to-end RE benchmark that spans from requirements elicitation to generation. We present Agent4RE - a self-refining multi-agent RE system that orchestrates specialized agents and incorporates two iterative improvement loops. To support evaluation, we construct RE-E2E - a real-world dataset built from human-written requirement specifications, enabling end-to-end assessment of RE workflows. Building on this foundation, we further propose two enhanced Agent4RE versions that incorporate either autonomous self-refinement or structured human feedback, and analyze their strengths and limitations across different scenarios. Evaluation on 8 Large Language Models (LLMs) demonstrates that all three Agent4RE variants consistently outperform a domain-context-augmented prompting baseline by average 8% in text-based metrics. The two enhanced variants achieve the highest LLM-as-a-judge and human ratings, surpassing two RE baselines by approximately 0.8 points on a four-point scale. This consistent performance establishes Agent4RE as a practical end-to-end RE solution for industrial environments.

---

### Sample-Efficiency of Kolmogorov-Arnold Networks

Authors: Kevin Riehl, Shaimaa K. El-Baklish, Fan Wu, Anastasios Kouvelas
Published: 2026-10-07
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.10627v1

Deep reinforcement learning has achieved substantial performance gains over classical control approaches. Yet, a central challenge to learning in real-world applications is acquiring costly samples. Kolmogorov-Arnold Networks are a recently proposed architecture that can learn physical relationships in control problems effectively, with significantly higher parameter efficiency and interpretability when compared to Multi-Layer-Perceptron architectures. In this work, we systematically study sample-efficiency using computational experiments, covering the Feynman dataset and the Gymnasium RL benchmark. The results show that similar performance can be achieved with 40% fewer samples using the Kolmogorov-Arnold architecture, and that relative performance improvements up to 50% occur during the training process. The observed gains are robust to varying levels of noise in rewards. These results highlight the potential of the Kolmogorov-Arnold architectures for more sample-efficient reinforcement learning. Code: https://github.com/DerKevinRiehl/neurips26_kan_training

---

### Exact SO(3)-Equivariant Isotropic Kernels for Rotation-Robust Neural Dynamics

Authors: Ridham Patel
Published: 2026-10-07
Categories: cs.LG, cs.CE, stat.ML
arXiv: https://arxiv.org/abs/2610.10626v1

Neural surrogates for vector-valued partial differential equations can fit training data yet change their predictions when the same physical state is expressed in a rotated coordinate frame. We study this failure on three-dimensional Navier--Stokes dynamics observed at irregularly placed points. We introduce the Invariant-Conditioned Isotropic Kernel Neural Operator (IKNO), a compact graph model that builds local interactions from scalar quantities unchanged by rotation and vector directions that rotate with the data. Consequently, rotating the positions and velocities rotates the predicted velocity change in exactly the same way. On a held-out test set fixed after model design, training unconstrained graph models on randomly rotated examples reduces but does not eliminate their coordinate dependence. In contrast, IKNO is consistent to numerical precision, matches the forecasting accuracy of a general rotation-aware Tensor Field Network with $5.6$ times fewer parameters, and outperforms a parameter-matched graph simulator. These results show that a compact, PDE-specialized model can remove coordinate dependence without sacrificing forecasting accuracy.

---

### Safe at One Loop, Risky at Another: Aligning Safety Across Recurrent Depths in Looped Language Models

Authors: Yi Wang, Xiuyuan Qi, Dongqi Han, Dongsheng Li, Wenjie Wang
Published: 2026-10-07
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2610.10625v1

Looped Language Models (LoopLMs) provide a parameter efficient approach to scaling model capabilities through repeated use of shared parameters across recurrent steps. Since each recurrent depth can be read out independently, a single LoopLM exposes a broader output space across inference depths, raising an important question: whether safety is preserved throughout recurrent computation. Prior evaluations suggest that deeper recurrence can improve safety on harmful queries, but robustness under jailbreak attacks remains unclear. We therefore conduct a comprehensive safety evaluation of LoopLMs under jailbreak attacks targeting different recurrent depths. We find that attack success can increase at deeper inference depths, the same query can elicit different safety behaviors across depths, and attacks constructed against one depth can transfer to others. Moreover, SFT and preference alignment do not eliminate these safety gaps, motivating an alignment method designed for LoopLMs. We introduce SafeBridge, which combines lightweight depth specific control of shared recurrent layers, selective state bridging, and joint safety supervision across recurrent depths. Across model scales, multiple attack methods, and safety benchmarks, SafeBridge substantially reduces attack success for both matched-depth and cross-depth attacks. It also improves general utility over the vanilla models while maintaining comparable over-refusal behavior. Our results show that the safety of a LoopLM cannot be inferred from a single recurrent depth, motivating safety alignment across recurrent computation. Our code and model checkpoints will be released upon acceptance.

---

### SAVU-BENCH: A Real-World Benchmark for Spatial Audio-Visual Understanding

Authors: Yu Chen, Ruihang Liu, Yangguang Xu, Xinyue Jiang, Mohammed Bennamoun, Farid Boussaid, Xinyuan Qian, Qiuhong Ke
Published: 2026-10-07
Categories: cs.SD, cs.AI
arXiv: https://arxiv.org/abs/2610.10624v1

Spatial audio-visual understanding requires models to recognize not only what is present, but also where events occur and how they relate across modalities. Existing benchmarks often rely on simulated scenes, evaluate isolated spatial skills, and provide limited diagnostic insight into failure modes. We introduce SAVU-Bench, a real-world benchmark that systematically evaluates spatial audio-visual understanding across three capability levels and seven evaluation tasks. We further introduce SAVU-Diag, a scene-linked diagnostic set that decomposes reasoning questions into their prerequisite grounding and alignment sub-tasks. Evaluation of 12 representative models on SAVU-Bench reveals that while visual spatial grounding is relatively mature, spatial perception involving audio remains a primary bottleneck. SAVU-Diag further demonstrates that most reasoning errors co-occur with failures on these prerequisite tasks, though reasoning gaps persist even when prerequisites are correctly resolved. Motivated by these findings, we introduce SAVU-EA, a training-free evidence-augmented baseline that makes spatial cues more explicit. While SAVU-EA substantially improves spatial grounding and joint matching, high-level spatial reasoning remains challenging. Our findings highlight the urgent need for both robust spatial audio perception and deeper integration of cross-modal spatial relations.

---

### Recurrent Self-Improvement: Dynamic Cross-Loop On-Policy Distillation for Looped Language Models

Authors: Yi Wang, Rui Qian, Yu Li, Haoyang Yao, Wenjie Wang
Published: 2026-10-07
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2610.10623v1

Looped Language Models (LoopLMs) offer a parameter efficient approach to scaling reasoning by reusing shared parameters across recurrent computation steps. Despite their promise, effective post-training of LoopLMs remains challenging. Existing approaches either provide reward based supervision that is sparse or costly to extend across loops, or rely on external teachers or privileged information, leading to limited teacher availability or teacher-student context mismatch. To address these limitations, we introduce LoopOPD, a cross-loop on-policy distillation framework that uses additional recurrent computation within a LoopLM as its own source of supervision. LoopOPD uses a frozen terminal loop policy as a compute privileged teacher for an intermediate loop student on student generated rollouts, providing dense supervision without an external teacher or privileged information. We further propose Dynamic LoopOPD (D-LoopOPD), which continually refreshes the terminal loop teacher as the shared model parameters are updated, enabling recurrent self-improvement. We characterize how distillation updates propagate across loop depths and derive sufficient conditions under which a single update yields simultaneous local improvement at both loop depths. Experiments on Ouro-Thinking models show that LoopOPD improves mathematical reasoning, while D-LoopOPD yields further gains through dynamic teacher updates. Despite being trained only on mathematical data, the resulting models also improve on general reasoning and code generation benchmarks, demonstrating that recurrent computation can serve as an effective source of supervision for LoopLMs. Our code and model checkpoints will be released upon acceptance.

---

### WorldBench: Evaluating LLMs on Three.js Voxel World Generation

Authors: Krish Bakshi
Published: 2026-10-07
Categories: cs.GR, cs.AI, cs.CL, cs.CV, cs.SE
arXiv: https://arxiv.org/abs/2610.10622v1

Large language models can now write complete, interactive 3D worlds as code, but grading those worlds automatically is unreliable. Existing judges take one view of the output: a vision-language model scores a few rendered snapshots, or a language model reads the source. On worlds written by five frontier models we find that the two views disagree on 32% of required items, mostly code that no frame shows, and that fixed views miss small close-up contents. We present WorldBench, a benchmark and judge for open-ended, LLM-generated Three.js worlds. From one prompt describing a floating voxel island with ten biomes, physics, and day/night and seasonal cycles, the judge explores the running world, controlling its clock, orbiting it, and sending a navigator agent to frame each biome, and reads the code for what it sees. Neither channel is trusted on its own: a code quote counts only if it is text the source contains, and visual claims are checked against measured pixels where the property is measurable. A mutation test, in which we remove features by construction, shows that code-only judging gives full credit to four of five removed features, because their code remains in the file. Our judge cuts the points kept on removed features by a third (5.44 to 3.55 of 7.11), and what it still credits is mostly code that exists but never runs. We evaluate five frontier models: Claude Fable 5.1, GPT-6 Astra, Kimi K3, Grok 4.7 and Gemini 3.1 Pro. Code, prompt, tests and judge configuration are available at https://github.com/KrishBakshi/worldbench

---

### Self-Organization from Constrained Geometric Radiation

Authors: Ming Lei
Published: 2026-10-07
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.10621v1

How does dynamic order emerge spontaneously in closed systems without external driving? Existing paradigms all require external energy flows, temperature quenching, or slow driving. Here we report constraint-induced self-organization via geometric radiation in coupled metric evolution systems. Simulations reveal a universal four-stage cycle: stress accumulation, super-exponential radiation, chaotic collapse, and convergence to a fractal limit cycle, a novel attractor topology we term the wedge-shaped attractor, with five quantized curvature states and fractal micro-fluctuations. We identify four jointly sufficient conditions: an irreversible geometric horizon, persistent stress injection from quantum coherence, endogenous geometric tension between incompatible curvatures, and effective fluctuations. Their synergy triggers a critical avalanche at the horizon boundary. We prove three theorems: the Geometric Horizon Theorem, the Geometric Energy Dissipation Theorem (implying wave-like entropy evolution in closed systems), and the Radiation as Phase Transition Channel Theorem. We further establish the Constraint-Induced Self-Organization Theorem: these conditions guarantee the complete cycle with probability one. Systematic scans reveal a critical noise threshold and power-law scaling of radiation onset. We verify universality across 12 configurations, multiple noise types, and three geometric flows. This work establishes a new paradigm for closed-system self-organization, forging an exact mathematical duality between classical nonlinear constraints and gravitational horizons.

---

### When AI Finds Hidden Messages, Does It Report?

Authors: William Guey, Rashik Jahangir, Pierrick Bougault, Vitor D. de Moura, Wei Zhang, José O. Gomes
Published: 2026-10-07
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2610.10620v1

When an assistant encounters a message for another AI, does it tell its user? Four fixed model-provider deployments perform simulated source tasks in 1,280 ordinary-note and 128 enhanced-note sessions. Harmless and harmful messages have matched plaintext and ROT13 versions, with no-message controls. Observers receive no decoder or decoded meaning; a requested reference code incentivizes inspection. Asking for reports increases rule-detected notifications identifying another AI as recipient by 53.1 percentage points for harmless ROT13 messages and 54.7 for harmful ones. This is a joint inspection, recognition, and notification effect; missing-response bounds are 38.3--77.3 and 36.7--78.1 points. Model-based trace checks identify eleven ordinary plaintext cases where agents interpret the message but do not notify their user. Seven encoded omissions are verified with enhanced notes; ordinary encoded omissions remain unverified. Seven simulated filename disclosures coexist with accurate review-status answers, and two answers use a planted false count. Interpretation, notification, and authorized task performance are distinct outcomes.

---

### TestJack: Should you trust the results in coding benchmarks? Agentic Coding Benchmarks Auditing via Evaluator Evolution

Authors: Shuangjie Yao, Hao Wang, Koushik Sen, Simin Chen, Baishakhi Ray, Dawn Song
Published: 2026-10-07
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2610.10619v1

Large language model (LLM) agents are rapidly reshaping software engineering, accompanied by an explosion of new code benchmarks. Yet nearly all existing benchmarks still rely on the same decades-old criterion: a solution is correct if it passes a fixed set of unit tests. Such tests are often insufficient: they check only part of what the task requires, so agents can reward hack them or silently miss required behavior while still passing every test. As a result, higher benchmark scores may partly reflect better adaptation to the evaluator rather than better problem solving. Existing works focus on static test augmentation: they strengthen each task's tests once, before any trial is seen, and thus overlook how real trials actually fail. We introduce TestJack, a scalable framework for evaluating patches beyond fixed tests. For each trial, TestJack generates tests targeting prompt requirements the patch may violate, retains only tests passed by the ground-truth patch, and re-examines any trial failures. Each confirmed failure is thus supported by a replayable test. To reduce evaluation cost, we also introduce a lightweight variant which audits a random sample of trials in depth and reuses the resulting tests across all trials for the same task. Across 6 frontier model backends and 5 benchmarks such as DeepSWE and SWE Marathon, we find that about 34.4% of the model trials currently judged correct violate the task requirements, lowering the overall resolution rate from 50.6% to 33.2%. Our results reveal a fundamental limitation of current coding-agent evaluation: as LLMs become better at optimizing against fixed evaluators, those evaluators themselves must become more adaptive.

---

### MRCert: Towards Post-deployment Patch Robustness Certification for Adversarially Patched Samples via Type-specific Masking

Authors: Qilin Zhou, Zhengyuan Wei, Haipeng Wang, Zhuo Wang, Shuo Liu, W. K. Chan
Published: 2026-10-07
Categories: cs.CR, cs.AI, cs.LG, cs.SE
arXiv: https://arxiv.org/abs/2610.10617v1

In post-deployment time, inputs to deep learning models may or may not be adversarially patched. Patch robustness certification on such inputs within a patch bound can verify their label benignity and should retain high prediction accuracy. However, existing smoothing-based and masking-based recovery defenders cannot achieve both simultaneously: they degrade the prediction accuracy much and cannot verify the benignity of the returned label of an adversarially patched input, respectively. We propose MRCert, the first masking-based certified recovery defender that shows the feasibility of achieving both. Unlike all existing works to apply a common condition across both types of input (benign and adversarially patched samples) for certification, MRCert infers type-specific necessary properties of deep learning models for both types in post-deployment time and formally relates them to verify the label benignity through a novel type-oriented design of label recovery and certification function pair. Without incurring the degradation in clean accuracy caused by smoothing, experimental results confirm that MRCert achieves 35.1\% adversarial certified accuracy on ImageNet at patch size 16 pixels, whereas the SOTA PatchCURE fails completely.

---

### When Routing Reveals Membership: Privacy Leakage from MoE Router Telemetry

Authors: Yixin Tan, Jiayang Liu, Lu Sun, Yuke Hu, Zheng Li, Rui Wen
Published: 2026-10-07
Categories: cs.LG, cs.CR
arXiv: https://arxiv.org/abs/2610.10616v1

Mixture-of-Experts (MoE) language models produce routing information during inference that may be logged or exposed for monitoring, debugging, load analysis, and safety auditing. Unlike ordinary model outputs, this telemetry reveals a view of the model's internal computation, raising a privacy question: can it reveal whether an example was used to fine-tune the deployed model? We introduce a router-augmented membership inference attack that combines conventional output-side signals with aggregated routing features and applies a membership classifier learned from independently fine-tuned shadow models to the target model. Across three MoE architectures and three data domains, router telemetry consistently improves membership inference over a strong output-signal ensemble, increasing TPR at 1\% FPR by 2.7--9.4 percentage points across all nine settings. The leakage persists across full fine-tuning, frozen-router training, LoRA, and instruction tuning, and remains observable with only discrete expert selections, restricted telemetry, or a single shadow model. Mechanistic analysis further shows that the leakage does not require router-specific memorization: fine-tuning introduces membership information into hidden representations, while the router exposes a projection of this signal even when its parameters are frozen. Perturbing the telemetry reduces this additional leakage only as its fidelity degrades. Our results show that router telemetry can turn an operational signal into an additional privacy surface for fine-tuned MoE models.

---

### JevForest: Path Voting for Budgeted Feature Acquisition

Authors: Yu Yan
Published: 2026-10-07
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2610.10615v1

Choosing which information to observe is central to prediction under limited observation budgets. We study JevForest, a feature acquisition policy that aggregates path-dependent proposals from bootstrapped trees, weights them by global training information gain, and predicts from the acquired values with a shared masked classifier. An online implementation queries Jev for semantic answers selected by this policy. On small balanced held-out samples, four-question forest acquisition achieves accuracy $0.729$ on AG News ($n=48$), compared with $0.667$ for a static gain ranking and $0.583$ for random ordering. On TREC ($n=24$), the ordering reverses: forest accuracy is $0.667$, compared with $0.750$ and $0.833$. Asking all eight questions in one batch yields higher accuracy at lower measured cost and latency than four sequential forest queries; direct Jev classification matches the batch accuracy while costing less. Offline MiniBooNE experiments yield accuracy $0.845\pm0.010$ at ten features and $0.885\pm0.008$ at forty features over three jointly varying data and forest seeds (mean $\pm$ sample standard deviation). A companion Newton boosting implementation provides preliminary full-feature synthetic results. These exploratory findings establish a working Jev acquisition workflow but do not support a general advantage for path voting: its value depends on the task, predictor, and the distinction between question budgets and actual query costs.

---

### Temporal transformer CAN encoder with federated lightweight heads for anomaly detection

Authors: Konstantinos Gyftodimos, Kyriakos Chiotis, Elena Politi, George Dimitrakopoulos, Eirini Liotou
Published: 2026-10-07
Categories: cs.LG, cs.CR
arXiv: https://arxiv.org/abs/2610.10613v1

Modern vehicles rely on large numbers of Electronic Control Units (ECUs) that constantly exchange information over the Controller Area Network (CAN) bus. Due to the rapidity, structure, and repetition of this communication, even slight variations in timing, payload values, or message patterns can point to unusual activity. Whether due to errors, malfunctions, or deliberate interference, these anomalies are frequently subtle and challenging to identify with conventional methods that handle messages separately or rely on manually created rules. Motivated by this gap, we present a privacy-preserving framework for anomaly detection in in-vehicle networks, based on a Temporal Transformer CAN Encoder with Federated Lightweight Heads, to better capture these irregularities. The detection of subtle temporal and contextual anomalies is made possible by a lightweight Transformer encoder that learns how these signals evolve over time, while a federated learning mechanism enables several vehicles or ECUs to work together to improve a shared model without exchanging raw CAN data. This combination of federated learning and temporal sequence modeling provides robust anomaly detection performance while maintaining efficiency and privacy, according to experiments conducted on open-source datasets.

---

### Verification and Self-Improvement in Agentic AI: Foundations and Limits

Authors: Chien-Ping Lu
Published: 2026-10-07
Categories: cs.AI, cs.CC
arXiv: https://arxiv.org/abs/2610.10611v1

Agentic AI systems can improve by searching longer, receiving additional support, or modifying how they propose and verify outputs. A performance score does not distinguish these mechanisms. We compare these changes through bounded verification with hidden terminal randomness. A stage specifies admissible transcripts, polynomial bounds, an alternating verification protocol, and a terminal checker. Its native reach uses default support; its closure frontier permits all support already admitted by the interface. Under a uniform pointwise probability gap and task-relative soundness, these are well-defined languages. We prove that independent majority amplification preserves both languages, whereas existential acceptance over random tapes can admit incorrect outputs. Exact verification is the zero-randomness case, with placement and completeness results. The randomized-verifier classes satisfy $Σ_k^{\mathrm{P}}\subseteqΣ_k^{\mathrm{RV}}\subseteqΣ_{k+1}^{\mathrm{P}}$; strict enlargement and depth separation require explicit complexity assumptions, while $\mathrm{BPP}=\mathrm{P}$ yields exact companions with the same frontiers. Representation analysis separates invariant acceptance from core-versus-support labels that can change under refactoring. For recursive self-improvement, uniformly bounded self-modification under a common sound interpreter and fixed verification protocol remains within the same verification class. A separate conditional-error budget controls false selection across adaptively chosen candidates. A quota-enforced XOR-synthesis family separates unbounded ratios of search success from changes in the accepted languages; exact and probabilistic audits check the resulting evidence requirements. The framework ties self-improvement claims to obligations on correctness, admissible evidence, verification resources, and selection error.

---

### Code Understanding is a Bottleneck for Coding Agents

Authors: Nishant Balepur, Kiran Tomlinson, Tobias Schnabel
Published: 2026-10-07
Categories: cs.SE, cs.AI, cs.PL
arXiv: https://arxiv.org/abs/2610.10610v1

Repository benchmarks (e.g., SWE-bench) for coding agents often assume that lines of code edited can predict task difficulty, but such datasets' poor control over code and task types makes it hard to know which abilities truly drive agent errors. We present CABRA: a Coding Ability Blueprint for Rigorous Agent evaluation. CABRA builds tasks from scratch as call graph transformations and scales difficulty via a task size parameter on four axes: function traversal, search, runtime resolution, and instruction following. We run eight LLMs and six coding agents on 6,840 CABRA tasks to show: 1) LLM accuracy falls as task size~grows, but agents stay near-perfect by offloading work to tools (e.g., grep); 2) Larger CABRA tasks elicit more tool calls for reading and analysis, while a separate study on SWE-bench Verified shows these tool call counts predict agents' accuracy better than lines of code edited, suggesting task difficulty for agents can lie in understanding code to edit, not just in making edits; 3) Extending CABRA to an intense understanding task where models analyze divergent logic across two classes backs this finding, as agent accuracy finally falls. More broadly, we argue for synthetic evaluations like CABRA to unmask LLM weaknesses trivialized by tools (e.g., needle-in-a-haystack) and abilities beyond just editing (e.g., understanding) that coding agents still find difficult, pairing SWE-bench-style tasks with controlled diagnosis.

---

### From Investigation Failures to Reliable SOC Agents: Understanding and Improving LLM-Based Alert Triage

Authors: Saimon Amanuel Tsegai, Alex Kantchelian, Danfeng, Yao, Peng Gao
Published: 2026-10-07
Categories: cs.CR, cs.AI, cs.MA
arXiv: https://arxiv.org/abs/2610.10608v1

Security operations centers (SOCs) must triage large volumes of alerts, most of which are benign, while missed attacks can remain uninvestigated. Tool-using large language model (LLM) agents can retrieve evidence during triage, but it remains unclear how reasoning strategies determine what to gather and when an investigation is sufficient to close an alert. We study five representative approaches spanning single-pass tool use, iterative retrieval, sampled investigations, self-review, and explicit verification. To support this study, we build ALERT-BENCH, an interactive benchmark that replays enterprise telemetry through a live SIEM and requires each system to retrieve evidence. Across 1,247 alerts from a multi-stage attack scenario, every approach missed at least 40.4% of attack-related alerts. Trace analysis shows that attack alerts are more likely to be dismissed when searches return no records, same-context review has negative net correction, and dismissal receives no consistently stronger investigation than escalation. Based on these findings, we further design AIDA (Adversarial Investigation and Dialectical Analysis), a multi-agent framework that requires an explicit proposed decision before independent challenge and stronger evidentiary requirements before dismissal. AIDA preserves investigation history in an append-only Investigation Ledger and keeps the challenge in a separate reasoning context. A separate Judge adjudicates the proposed decision and challenge against evidence, resolving the alert or requesting another round when evidence is missing. On the same alerts, AIDA achieves an F1 score of 0.958, compared with 0.371-0.744 for the studied approaches, and reduces the false-negative rate from 40.4% to 3.1% while escalating 18.4% of alerts to analysts. These results show that structuring evidence retrieval and decision review can substantially improve agentic SOC triage.

---

### A Camera-Native Stereo VR180 Dataset

Authors: Linxuan Lu
Published: 2026-10-07
Categories: cs.CV, cs.MM
arXiv: https://arxiv.org/abs/2610.10607v1

Immersive VR180 video is increasingly produced with professional stereo fisheye cameras, yet public VR180 research resources are mostly collected from online platforms such as YouTube: already stitched, projected and compressed by unknown pipelines, and without lens calibration. We present a firsthand-captured stereo VR180 dataset recorded with two Blackmagic URSA Cine Immersive cameras. It contains 1,211 samples -- 636 stereo video clips (2,220.8 s, mostly 90 fps) and 575 stereo stills -- each released as camera-native Blackmagic RAW, separate-eye native fisheye HEVC (8160x7200 per eye) and half-equirectangular HEVC (7200x7200 per eye), together with the factory lens calibration, portable fisheye/half-equirectangular conversion tools and AI-generated scene and visual-challenge annotations. Re-encoding the released fisheye and half-equirectangular renders with x265 over 24 clips, both eyes, four rate points and nine viewing directions, native-fisheye coding needed more bitrate than half-equirectangular coding at equal viewport quality for all 24 clips (median +38%), in every part of the field of view. Data: https://huggingface.co/datasets/lulinxuan/VR180 ; code: https://github.com/lulinxuan/vr180-dataset-tools

---

### Beyond Type-checking: Towards Holistic Evaluation of Formal Specification Generation

Authors: Srijith Nair, Aditya Vempaty, Jia Liu, Ashish Jagmohan
Published: 2026-10-07
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2610.10604v1

When generating verifiable code, natural language requirements are mapped to machine checked code using LLMs and agentic workflows. A crucial component of this pipeline is specification generation (SpecGen), which produces a formal contract against which an agent can prove implementation correctness. Proof generation can obtain deterministic feedback from a theorem prover, but SpecGen lacks a definitive check that a generated specification captures the user's intent. A checked proof can therefore establish correctness against a specification that misrepresents the intended behaviour. We take a step towards holistic SpecGen evaluation with a unified dataset assembled from $350$ existing Lean tasks, including $189$ from VERINA and $161$ from CLEVER, and a framework covering formal validity, reference similarity and equivalence, and behavioural adequacy. We distinguish acceptance of required inputs from acceptance of valid outputs and rejection of invalid outputs, while making each metric's evidence scope explicit. Across four SpecGen configurations, restricting the generalized tree edit distance (GTED) comparison, a reference similarity measure, to $32$ jointly measurable VERINA tasks changes the VERINA configuration's position from second to fourth in mean similarity, showing the importance of measurement coverage. In an authored control, a specification achieves $100\%$ positive test recall and negative test rejection while accepting $0\%$ of required inputs. This demonstrates that perfect postcondition scores can miss an unusable input contract, motivating separate feedback on input coverage and output constraints.

---

### Tetris3D: 3D Scene Generation With Objects That Fit Together

Authors: Jaeyeong Kim, Jinhyuk Jang, Jongmin Lee, Kyehong Park, Seungryong Kim
Published: 2026-10-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.10539v1

We propose Tetris3D, a generative framework for single-image 3D scene reconstruction that recovers objects which are physically and geometrically coherent as a scene. Existing methods often generate objects independently or couple them implicitly, providing limited guidance for ensuring fine-grained spatial compatibility between neighboring objects that interact with one another. To address this, we explicitly condition the generation of each object on the geometry of surrounding objects and their physical relationships, guiding its shape and pose to remain geometrically and physically plausible within the scene. Moreover, we introduce ComOb, a physics simulation-based dataset of 1.2M scenes featuring physical interactions across diverse object categories, with per-object meshes and pairwise physical relation annotations. Comprehensive experiments on synthetic and realworld scenes show that Tetris3D recovers coherent object shapes and poses even when interacting regions are occluded, and achieves state-of-the-art performance in both generation quality and physical stability.

---

### Never Look Back: Understanding Persistence in 3D Object Memory from Egocentric Videos

Authors: Shravan Chaudhari, William Paul, Suchi Saria, Rama Chellappa, Homanga Bharadhwaj
Published: 2026-10-07
Categories: cs.CV, cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2610.10538v1

As we move through the world and carry out everyday tasks, we encounter objects that may become relevant only later. We are capable of recalling where we left something or what was inside a container, even without knowing we would need it later. Here, we study how an embodied assistant can build a similar memory from egocentric videos, by observing a person's day-to-day activities. We present Ledger, a persistent 3D object memory that combines object locations, their histories, and contextual descriptions. It associates observations across the recording and retains objects after they leave the view, including those the person never touches. It clusters each object's observations by resting locations and records a move only after repeated evidence, reducing the effect of localization noise. Short descriptions preserve details such as an object's contents or supporting surface. It saves these records to later answer spatial questions without having to access the original images or video. Our memory raises HD-EPIC accuracy from 29.7% to 42.6%, UCS-Bench accuracy from 33.8% to 38.5% and localizes Ego4D objects with a 0.99 m median error on returned predictions. Our analyses identify complementary roles for temporal persistence, contextual descriptions, and retrieval. Our study on 100 stitched streams of multiple scenes each further exposes failures in both retrieval and construction. Per-scene construction partially recovers the performance lost across scene changes compared to that of single scene streams.

---

### Decoupling Exploration from Optimization in RLVR

Authors: Saif Punjwani, Micah Goldblum
Published: 2026-10-07
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.10536v1

Modern language models undergo reinforcement learning with verifiable rewards (RLVR) on top of already-trained checkpoints. A key promise of RLVR is the discovery of new reasoning strategies. In principle, a model can sample novel ideas absent from its prior training data. In practice, however, augmenting RLVR with strong novelty incentives has seen limited success and can degrade model quality. Because verifiable rewards supervise only a narrow slice of the model's knowledge and behavior, such degradations are difficult to recover from. Instead, we decouple exploration from optimization in a framework we call Exploration-Distillation (ExpDis). We train one or more explorer policies with a novelty bonus in the reward, filter their trajectories for correctness and quality, and distill them into a separate student policy. The student policy is then trained without a novelty bonus. We repeat the above procedure for several rounds, alternating between exploration and optimization. This decoupling allows us to aggressively scale exploration without degrading the student policy. Across seven mathematical reasoning benchmarks and two model families, ExpDis outperforms DAPO at the same wall-clock budget. Moreover, we observe improved pass@$k$ scaling, indicating that ExpDis produces models that generate more diverse correct solutions.

---

### EngramEdit: Decoupled Knowledge Updates in LLMs through Conditional Memory

Authors: Hongru Cai, Ran Wei, Wenjie Wang, Chengfa Wu, Ning Song, Yongqi Li, Wenjie Li
Published: 2026-10-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.10533v1

Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding the capacity of large language models (LLMs) with limited additional computation. Beyond model scaling, this architecture has demonstrated the potential to decouple factual knowledge storage from general-purpose computation, offering a promising route to updating factual knowledge while keeping the Transformer backbone fixed. Realizing this potential is challenging because different expressions of a fact may activate different n-gram embeddings, while updating shared embeddings can unintentionally change the model's predictions about other facts. We propose EngramEdit for decoupled knowledge updates through conditional memory. EngramEdit first computes target memory representations that make the model predict the updated fact across multiple expressions. It then jointly updates the shared n-gram embeddings to match these targets across expressions and edits, penalizing updates to frequently reused embeddings more strongly to preserve unrelated knowledge. Experiments show that EngramEdit enables independent factual knowledge updates through conditional memory, achieving near-perfect editing success. Revised knowledge is usable across unseen expressions and in multi-hop reasoning, with nearly three times the strongest baseline's accuracy under chain-of-thought (CoT) prompting. Unrelated knowledge and general capabilities are largely preserved even as factual updates accumulate. These findings show that EngramEdit turns conditional memory into an editable knowledge interface, extending its role beyond model scaling to support decoupled knowledge updates.

---

### Long-WAM: Scaling the Context of World-Action Models

Authors: Wei Huang, Bohan Zhang, Chenzhi Liu, Isabella Liu, Shuai Yang, Weian Mao, Luozhou Wang, Yicheng Xiao, Weifeng Lin, Qixin Hu, Bryan Chu, Sifei Liu et al.
Published: 2026-10-07
Categories: cs.RO, cs.AI, cs.CV
arXiv: https://arxiv.org/abs/2610.10528v1

Real-time robot control demands enough visual history to infer motion and task progress, but processing that history can delay action. We present Long-WAM, a model-system framework for scaling the context of causal world-action models under real-time control constraints. Our central finding is that access to history is not the same as using it: longer histories pay off far more when the video foundation is pretrained autoregressively (AR). We first learn causal prediction from robot and egocentric videos without action labels, then preserve this history-to-future structure during world-action adaptation. On RoboCasa GR-1, increasing context from 0.0 to 19.2 seconds raises success from 63.3% to 78.7%, whereas a bidirectionally pretrained initialization shows no net gain; robot-domain AR pretraining further raises peak success on GR-1 and LIBERO-Long. Long-WAM also achieves the best results among compared methods on LIBERO-Long, RoboTwin 2.0, and DOMINO. Streaming observation encoding, asynchronous execution, and hardware-specific acceleration enable deployment on RTX 5090, DGX Spark, and Jetson AGX Thor without dropping future prediction; on RTX 5090, each action chunk, including future-video latent prediction, takes 107.4 ms. Real-time deployment on Unitree G1 and YAM supports dynamic and long-horizon manipulation, including 95% success on dynamic cup stacking, where Pi0.5 and Fast-WAM succeed in none of 20 trials. As a memory-informed executor, Long-WAM also complements higher-level planning in composite tasks.

---

### Decentralized SGD under Heavy-Tailed Noise: Optimal Convergence Rates and the Role of Gradient Clipping

Authors: Aleksandar Armacki, Haoyuan Cai, Ali H. Sayed
Published: 2026-10-07
Categories: math.OC, cs.LG, cs.MA
arXiv: https://arxiv.org/abs/2610.10527v1

Heavy-tailed noise has been widely observed in modern machine learning, motivating the use of methods like gradient clipping and normalization. While these methods are well understood in centralized settings, much less is known in decentralized ones, where applying a nonlinearity to local gradients affects both optimization and consensus. Recent works on decentralized non-convex optimization have studied both clipping and normalization under heavy-tailed noise, with clipping yielding suboptimal rates and normalization needing local momentum or mini-batches to converge. This raises the question: can a baseline decentralized method using a nonlinearity achieve optimal convergence rates under heavy-tailed noise? We answer affirmatively with clipped decentralized SGD ($\mathtt{DSGD}$). For smooth non-convex costs under bounded $p$-th moment noise, $p \in (1,2)$, we show that clipped $\mathtt{DSGD}$ achieves order-optimal rates both with high probability and in expectation. Moreover, we establish a linear speed-up in the number of agents, which, to our knowledge, has not been shown for decentralized methods with clipping. The key technical ingredient is a sharp analysis of the consensus gap that exploits the structure of clipping, relegating network effects to higher-order terms. Our results highlight an important distinction between clipping and normalization in decentralized settings: while normalized $\mathtt{DSGD}$ can fail to converge, clipping retains magnitude information, enabling $\mathtt{DSGD}$ to be convergent and order-optimal. Numerical experiments validate our theory.

---

### Rephrase Before You Act: Characterizing and Mitigating Language Sensitivity in Vision-Language-Action Models

Authors: Mikey Watts, Yuchen Cui
Published: 2026-10-07
Categories: cs.RO, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2610.10526v1

Vision-language-action models (VLAs) are strikingly sensitive to instruction phrasing and do not inherit the language robustness of the vision-language models they are built on. A one-word edit can move success by tens of points: $π_{0.5}$ turns on a LIBERO stove 100% of the time for "switch on the stove" and 2% for "switch on the hot plate", and a $π_0$ checkpoint finetuned with rephrase augmentation still shows swings of up to 61 points. We characterize this sensitivity with statistically tested single-edit swings and an oracle phrase search, which shows that phrasing alone nearly closes the 21-point gap between in-distribution and out-of-distribution tasks. We then reduce it without modifying the policy. Because the sensitivity is systematic, it can be expressed as explicit rules: we score many phrasings of a few training tasks, have a large language model distill the evidence into ten to twenty rephrasing rules, and at deployment rewrite each incoming instruction once under these rules. The rules improve the frozen $π_0$ by 16 to 27% relative on twelve held-out tasks across adversarial, VLM-generated, and human-generated phrasings, with gains concentrated on out-of-distribution tasks. The pipeline replicates on $π_{0.5}$ and LIBERO, lifting in-finetune success from 93.6% to 97.8%. The method requires no retraining and no per-step verification, and applies zero-shot to unseen tasks and instructions. Project website: https://sttawm.github.io/rephrase-before-you-act

---

### GRACE: Generation-aware latent compression for efficient video generation

Authors: Jiyoung Kim, Paul Hyunbin Cho, Jisu Nam, Donghoon Lee, Hyunsung Go, Yeonkyeong Lee, Hansaem Kim, Seungryong Kim
Published: 2026-10-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.10524v1

Highly compressed video autoencoders offer an effective way to accelerate video diffusion models, as the Diffusion Transformer (DiT) operates on far fewer tokens. However, such autoencoders are challenging to train, since a higher compression ratio degrades reconstruction quality and recovering it requires more channels, which is known to slow the convergence of the DiT. The compressed latent also differs from the one the DiT was trained on, so the pretrained DiT must be either retrained from scratch or adapted at considerable cost. Compressing the autoencoder the DiT was trained with appears to preserve compatibility, yet optimizing it for reconstruction alone still shifts the latent away from the distribution the DiT has learned. To address this, we propose Generation-Aware Latent Compression for Efficient Video Generation (GRACE), a two-stage framework that compresses a pretrained video autoencoder while keeping it compatible with the pretrained DiT. Specifically, we keep a frozen base latent from the pretrained encoder and learn a residual latent for the information lost under stronger compression, while aligning the compressed latent with the pretrained latent in the feature space of the frozen DiT so that the autoencoder is optimized for generation. We then adapt the DiT with lightweight fine-tuning and asymmetric denoising, where the base is denoised ahead of the residual. GRACE reduces the token count of Wan2.1-I2V-14B by 8x and its latency by 11.1x at 480x832x81, while matching the generation quality of the pretrained pipeline before compression on VBench.

---

### Distilling Graph Geometry: Knowledge Gap from GNNs to MLPs

Authors: Zhewei Chen, Hao Zhu, Jiaojiao Jiang, Ahad N. Zehmakan
Published: 2026-10-07
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.10520v1

GNN-to-MLP distillation aims to retain the predictive accuracy of a message-passing teacher while deploying a graph-free MLP at inference. Existing methods mainly transfer node-wise predictions or use confidence-based reweighting, but they do not specify where the student should preserve the teacher's graph-induced geometry. We show that this omission leads to two spectral failure modes in the student's representation space. On sparse graphs, the student suffers from spectral underfit, missing high-energy teacher directions concentrated near boundary regions. On dense graphs, it suffers from spectral overfit, retaining spurious directions that the teacher has collapsed through aggregation. Motivated by an energy-weighted teacher-student alignment objective, we propose Graph Geometry-aware MLP (G^2MLP), a training-time distillation framework guided by Ollivier-Ricci curvature. Curvature identifies where the two spectral errors concentrate and is used to allocate supervision between prediction-level and representation-level alignment. The deployed model remains a standard MLP and requires no graph access at inference. Across node-classification benchmarks, G^2MLP consistently improves over graph-free distillation baselines, reduces the teacher-student rank gap in both regimes, and transfers without architectural changes to Graph Transformer teachers and link prediction.

---

### Why Forget-Only Unlearning Needs Memorization

Authors: Luka Radić, Vikrant Singhal, Amartya Sanyal
Published: 2026-10-07
Categories: cs.LG, cs.IT, stat.ML
arXiv: https://arxiv.org/abs/2610.10519v1

Machine unlearning asks for a deletion algorithm whose output is close to retraining from scratch without the selected forget examples. In this work, we study forget-only unlearning, where the deletion algorithm receives only the trained model and the examples to forget, with no retained data or extra training information. We ask whether forget-only unlearning is always possible. We first show that this depends on the learning method: different datasets can produce the same trained model but require very different outputs after the same examples are removed. Using this observation, we derive lower bounds on how accurately unlearning can match retraining and instantiate them for several standard learning algorithms. We then ask what must be true when forget-only unlearning succeeds. To this end, we derive lower bounds on what an algorithm must memorize about the training data to handle arbitrary deletion requests. For simple threshold learners, the required information can be as large as the entire dataset, even though ordinary training keeps only one boundary point. Overall, our results show that information discarded during ordinary learning may be needed later for deletion, so models designed for forget-only unlearning may need to retain more information than standard training does.

---

### RoboJEPA: Scaling Robotic Latent World Models

Authors: Artem Zholus, Nicolas Beltran-Velez, Jianhao Yuan, Sarath Chandar, Tushar Nagarajan, Daniel Severo, Koustuv Sinha, Michal Drozdzal, Adriana Romero Soriano, Jeannette Bohg, Nicolas Ballas, Mahmoud Assran
Published: 2026-10-07
Categories: cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2610.10515v1

Latent world models have shown a remarkable ability to predict future states and to plan in the real world. In practice, however, we lack a principled way to estimate how their capabilities scale with model size, data, and compute, an open problem that slows progress in the field. In this work we present RoboJEPA, a world model based on the Joint Embedding Predictive Architecture (JEPA) and trained on a large-scale dataset spanning 12 robotic embodiments. We show that RoboJEPA's imagination error, the error of its latent rollouts, follows a second-order power law in compute, allowing us to predict model quality well beyond the scale at which the law is fit. We further show that downstream robotic planning performance improves predictably with compute, and that imagination error is strongly correlated with it, making it a reliable proxy for real-robot evaluation. Finally, we demonstrate that latent world models can be deployed zero-shot as robotic agents, planning toward a single goal image to solve tasks requiring long-horizon planning on real hardware. We release all model checkpoints together with our training and robot deployment code. To our knowledge, this is the first work to establish scaling laws for multi-embodiment robotic world models trained on real robot data, and RoboJEPA, at 8B parameters, is the largest JEPA predictor model trained to date.

---

### SciExam for ENSO: Can AI Agents Build Climate Models?

Authors: Yinling Zhang, Langchen Liu, Dongbin Xiu, Xueyan Zou, Xu Kuang, Mengdi Wang, Shilong Liu
Published: 2026-10-07
Categories: cs.AI, cs.LG, physics.ao-ph
arXiv: https://arxiv.org/abs/2610.10513v1

Language-model agents are increasingly asked to carry out open-ended scientific research, yet their results are usually graded against a known answer, a rubric, or a language-model reviewer, none of which can tell whether a new scientific model is valid. The AI Science Exam for El Nino-Southern Oscillation (SciExam for ENSO) is a benchmark in which agents build low-order stochastic models of ENSO, the dominant mode of interannual climate variability, from real observations. Within a six-hour budget, agents process the observations, write their own diagnostics, which are then frozen, and develop a model using only these diagnostics as feedback. Hidden graders then test whether the model reproduces ENSO's statistics, recovers unobserved variables, and forecasts held-out years, and score a published model in the same way. Across twelve agent systems, six produce models that score higher than the published model, mainly through better reconstruction and forecasting. The simplified forms of the stronger models are each compatible with one of the two competing explanations of ENSO's warm-cold asymmetry, an open debate that the task never mentions. Controlled runs of the top system under varied information suggest that its scores do not come from recalling the dated observational record and that the information it receives shapes how it builds its model. SciExam for ENSO can thus evaluate agent research where no answer is known, and the results suggest that agents can already build competitive models whose structures bear on questions that scientists still debate.

---

### Video-Conditioned Generative Joint 2D-3D Hand Motion Recovery

Authors: Chen Xu, Yunqi Li, Binbin Huang, Brent Yi, Shenghua Gao, Yi Ma
Published: 2026-10-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.10512v1

Recovering faithful 3D hand motion from video remains challenging due to frequent occlusions and incomplete visual observations, which make frame-wise pose estimates unreliable and temporally inconsistent. To address this problem, we propose JoHan, a unified generative framework that recovers hand motion directly from video sequences without relying on intermediate per-frame pose predictions. Trained from scratch, our model jointly generates aligned 2D and 3D local hand pose sequences by learning their temporal dynamics and cross-representation correspondence. The generated 2D trajectories exploit direct spatial and temporal cues from the 2D images to guide the following generative 3D motion reconstruction, while the learned motion prior promotes temporal consistency. Their learned 2D-3D correspondence further enables recovery of the hand's global position and orientation relative to the camera. Extensive experiments on challenging benchmarks demonstrate significantly improved accuracy and speed in local hand-pose and camera-space reconstruction. Notably, our method captures much better hand-motion dynamics, producing significantly smoother motion than previous methods while maintaining high per-frame pose accuracy.

---

### Your Prompt Should Do More: Effects of Retrieval Instructions in Embedding Models

Authors: Amanda Myntti, Jenna Kanerva, Veronika Laippala, Filip Ginter
Published: 2026-10-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.10508v1

Prompted embedding models have recently received increasing attention, particularly for retrieval, where detailed retrieval instructions are provided as part of the retrieval prompt. Several new datasets and studies have examined this setting, showing that the current embedding models often struggle to follow such instructions reliably. In this paper, we study the mechanism of how instructions actually affect the representations of retrieval queries in asymmetric retrieval tasks. We show that models can fail to follow even simple task instructions when query-side distractors are included in the evaluation. We hypothesize that this behavior is driven by the training setup of current embedding models and their evaluation, and show that fine-tuning with added query-side distractors leads to substantial improvements, with minimal effect on other tasks.

---

### RECAST: Learning to Compute the Right Context through Adaptive Evidence Routing

Authors: Yilun Hao, Krishna Sayana, Isabella Ye, James S Ren, Sukhdeep Sodhi, Craig Boutilier, Chuchu Fan
Published: 2026-10-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.10507v1

Large language models are increasingly applied to tasks grounded in long, heterogeneous information sources. Conventional Retrieval-Augmented Generation (RAG) relies on fixed similarity-based retrieval, while agentic variants adapt queries and tool use but remain largely retrieval-centric. However, in many tasks, the evidence required for a solution is not explicitly present in any single source item. Instead, it must be derived through filtering, aggregation, or computation across multiple source items. In this work, we introduce RECAST (Routing Evidence through Computation, Access, and Synthesized Tools), a learned framework that formulates evidence construction as a sequential decision process over heterogeneous retrieval and computation operations, allowing evidence to be actively derived rather than merely retrieved. A lightweight RouterLM iteratively selects and formulates primitive operations or specifies customized operations for a frozen CompilerLM to translate into executable code. Once it judges the evidence sufficient, RouterLM passes the accepted evidence to a frozen AnswerLM to produce the final solution. We train RouterLM with supervised fine-tuning (SFT) followed by group relative policy optimization (GRPO). Across six heterogeneous benchmark families, RECAST achieves a mean success rate of 75.6%, outperforming the strongest large-model baseline by 15.9%. Moreover, training enables the Qwen3.5-9B RouterLM to outperform a training-free Gemini 3.5 Flash RouterLM by 5.0%. On three held-out benchmarks, RECAST improves over the strongest baseline by 15.0% on average, demonstrating strong zero-shot generalization across tasks and heterogeneous source representations.

---

### Validity Without Ground Truth: What Stated-Preference Economics Offers the Evaluation of Language Models

Authors: Daniel Robert Kling Alexander, Catherine Louise Kling
Published: 2026-10-07
Categories: cs.AI, cs.CL, econ.GN
arXiv: https://arxiv.org/abs/2610.10506v1

Many of the questions now put to large language models have no correct answer to score against: what a policy is worth, which option a user should choose, how to weigh competing values. Stated-preference economics has faced this problem for decades. It judges survey responses without knowing the true value, through a framework of validity and related concepts: content, construct, and criterion validity, reliability, incentive compatibility, and consequentiality. We argue that this framework is a general method for evaluating language models, and we set out what each concept means for LLM evaluation. We demonstrate the approach using a published water-quality stated preference economic valuation survey (Vossler et al. 2023) administered to six models. In this economic application, the validity tests take the form of predictions from economic theory: demand should slope down, and willingness to pay should respond to the scope of the good and to income. The tests separate the models sharply. Two older models fail the most basic test at a household income level of \$75,000, and the two newest pass every test of theoretical validity we can score, but diverge on convergent validity. Passing validity tests shows that a model's answers are coherent, not that they are correct.

---

### Oracle-Efficient and Parameter-Free Agnostic Smoothed Online Learning

Authors: Sasha Voitovych, Adam Block, Alexander Rakhlin, Abhishek Shetty
Published: 2026-10-07
Categories: cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2610.10499v1

Online learning is an attractive framework in many domains because it permits well-defined learning even when data are dependent or chosen adversarially. This generality, however, comes at a steep price, introducing significant statistical and computational barriers. Recently, smoothed online learning has emerged as a promising framework that interpolates between the fully adversarial and fully stochastic settings by assuming that the conditional law of each covariate has density at most $1/σ$ with respect to some fixed base measure $μ$, and it is known to match the statistical and computational guarantees of classical learning while still allowing for much of the flexibility of online learning. However, existing oracle-efficient algorithms require either (i) sampling access to the base measure $μ$ or (ii) labels that are perfectly predicted by a fixed hypothesis. Both assumptions limit the applicability of these algorithms, in contrast to statistical learning, where empirical risk minimization (ERM) learns efficiently in the agnostic setting without any knowledge of the data distribution. We show that neither assumption is necessary, giving the first oracle-efficient algorithm that achieves sublinear regret in the agnostic setting without knowledge of $μ$. Our algorithm, based on Gaussian Follow-The-Perturbed-Leader, is parameter-free: it requires no knowledge of $μ$, the smoothing parameter $σ$, or the horizon $T$, and it achieves regret $\widetilde O(d\sqrt{T/σ})$ for binary classes of VC dimension $d$ with a single call to an ERM oracle per round, which is optimal up to a $\sqrt{d}$ factor. En route to establishing the regret bound, we introduce several new techniques that may be of independent interest.

---

### EmbodiedRSI: Active Continual Robot Learning Through Hypothesis-Guided Co-Evolution

Authors: Python Song, Zhixuan Liang, Kelsey Fu, Mengdi Wang, Junfeng Yang, Shilong Liu
Published: 2026-10-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.10498v1

Robot foundation models provide strong visuomotor control, yet their performance can degrade when object positions or task instructions change. Further improvements often require post-training on substantial robot data, which can be costly to collect through methods such as teleoperation. Agentic harnesses can adapt around the model, but current self-evolving harnesses use robot trials inefficiently when deciding which code and skill changes to pursue. We introduce EmbodiedRSI, a self-evolving agentic harness that autonomously decides where to explore next and turns the resulting physical interaction into improved code and skills. EmbodiedRSI realizes this through a Fast-Slow Dual-System Architecture, in which competing code and skill hypotheses are maintained in a Hypothesis Graph. Value-of-Information Experiment Selection chooses physical experiments that can distinguish these hypotheses. Their outcomes guide Code-Skill Co-Evolution. The Slow System builds Hierarchical Memory, and Reward-Grounded Memory Learning selects effective memory according to their value for later Fast-System improvement. On RoboCasa365, EmbodiedRSI reaches 77.0% overall success and 71.3% on Composite-Unseen, compared with 40.1% for the best baseline. EmbodiedRSI also reaches 86.8% overall success on LIBERO-Pro. Beyond benchmark performance, EmbodiedRSI transfers zero-shot to real-world robot, achieving 71.3% overall success across multiple challenging tasks.

---

### QuadTok: Quadtree Visual Tokenizer for Autoregressive Image Generation

Authors: Yucheng Mao, Zeyuan Chen, Xiaojun Shan, Xiang Zhang, Divyansh Srivastava, Bingnan Li, Zhuowen Tu
Published: 2026-10-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.10497v1

We introduce QuadTok, a novel framework for visual tokenization and autoregressive image generation. Compared to traditional approaches using 2D grids or 1D token sequences, we propose a hierarchical quadtree structure, bridging the gap between 2D spatial binding and 1D sequence-level flexibility. The QuadTok tokenizer dynamically allocates representational capacity to visually intricate areas while leaving homogeneous regions at a coarse resolution. Compared with a fixed 256-token grid, our ImageNet-trained tokenizer saves approximately 10% of tokens on ImageNet and 9% when transferred zero-shot to the COCO dataset, while maintaining comparable reconstruction fidelity. Furthermore, the natural causality introduced by the tree structure seamlessly enables autoregressive image generation. Conditioned on a quadtree topology supplied before generation, our 947M GPT-style generative model achieves a 2.08 gFID on the ImageNet $256 \times 256$ benchmark. Additionally, leveraging the strong spatial correlation preserved by the quadtree structure, the QuadTok generator enables zero-shot spatially controlled image generation capabilities. Code: https://github.com/myc634/QuadTok.

---