# Research Papers — 2026-09 · bundle page 35 of 266 > Full-text markdown of papers #1701–1750 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/35 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/34.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/36.md) ### Impact of Patient Orientation in Single- and Multi-View Camera Environments for AI-based Rehabilitation Monitoring Authors: Miriama Jánošová, Andreas Lang, Petra Budikova, Jan Sedmidubsky Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35726v1 Automated quality assessment of rehabilitation exercises relies heavily on accurate human pose estimation from video data. Although numerous RGB-based pose estimation methods have been proposed, the impact of camera placement on detecting clinically relevant movement errors remains insufficiently explored. To address this gap, we introduce REHAB26-ViewAngles, a dataset comprising correct and incorrect rehabilitation exercise executions captured from a wide range of camera angles. Furthermore, we propose a novel separability metric to quantify an algorithm's ability to distinguish between valid and faulty exercise repetitions. Using these tools, we analyze how various RGB-based pose-estimation strategies are suitable for exercise quality assessment under varying camera placements. In particular, we analyze single-camera 2D and 3D pose estimation and four multi-camera strategies: a combination of two orthogonal 2D views, 3D triangulation, weighted 3D fusion, and an AI-based pose-estimation transformer model specifically trained from two synchronized cameras. Our findings reveal that an optimally placed 2D camera can improve the separability by 16.9\,\% over the commonly used $0^\circ$ frontal view and frequently outperforms single-camera 3D estimation, while combining two views can further improve accuracy by up to 13.1\,\%. These results offer practical guidance for deploying rehabilitation monitoring in both home and clinical settings. --- ### Superquadric Primitive Decomposition of 3D point clouds via Geometric-Aware Inlier Refinement Authors: Alessandro Rinaldi, Edoardo Tedesco, Andrea Ferraris, Filippo Leveni, Daniele Baieri, Filippo Maggioli, Simone Melzi, Luca Magri Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35725v1 The decomposition of 3D point clouds into interpretable geometric primitives remains a longstanding challenge in Computer Vision and Computer Graphics. Among the available representations, superquadrics offer a compact and expressive model capable of capturing a wide range of shapes. However, their estimation is inherently challenging, as it requires solving a non-linear optimization problem and is particularly sensitive to noise, outliers, and overlapping structures. While robust estimation methods such as RANSAC and its variants achieve strong performance, they rely primarily on spatial proximity and residual-based criteria, often leading to incorrect inlier assignments across adjacent or complex arrangements of primitives. In this work, we introduce a geometric-aware framework for primitive decomposition that explicitly incorporates local surface properties into the fitting process. Specifically, we propose an inlier refinement step formulated as an energy minimization problem and solved via graph-cut optimization. Our formulation integrates geometric priors, such as normal consistency, enabling more reliable inlier selection beyond purely residual-based criteria. The approach naturally applies to both single-model estimation and multi-model decomposition. By leveraging geometric information beyond point-wise residuals, our method reduces erroneous inlier propagation and stabilizes parameter estimation. Experiments on synthetic and real datasets show consistent improvements in geometric accuracy, robustness to noise and outliers, and convergence efficiency compared to state-of-the-art RANSAC-based methods. --- ### Hard Vision, Easy Vision: What GPT-6 Astra Reveals Across Computer Vision Authors: Hanoona Rasheed, Mohammed Irfan Kurpath, Bin Ren, Hisham Cholakkal, Fahad Shahbaz Khan, Salman Khan Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35718v1 Frontier general-purpose systems are rapidly expanding beyond visual understanding into capabilities traditionally handled by dedicated computer-vision models. As these capabilities expand, a central question for the computer-vision community is how far this reach extends, and what remains hard. We evaluate GPT-6 Astra alongside five frontier general-purpose AI systems across 34 capabilities and 55 benchmarks spanning nine areas of computer vision. We compare their performance with dedicated models and humans where suitable references are available. Astra demonstrates broad visual capability, with substantial gains over other frontier systems in visual and spatial reasoning and several forms of structured prediction. Across the state-of-the-art systems, a consistent pattern emerges. Capabilities involving semantic interpretation, reasoning, and object-centric prediction increasingly approach or reach available reference levels. In contrast, larger gaps remain when tasks require metric geometric accuracy, faithful reconstruction, temporally consistent dense prediction, or specialized fine-grained visual knowledge. Additional reasoning and specialist tools close selected gaps, but their benefits vary across capabilities. These results map a changing landscape of computer vision in which increasingly sophisticated visual tasks are accessible through a general-purpose interface, while precise and fidelity-sensitive perception remains an important frontier. --- ### X-Reset: Scaling Object-Centric Reinforcement Learning via Cross-Embodiment Resets Authors: Prithwish Dan, Chenyang Ma, Wei Zhan Published: 2026-09-28 Categories: cs.LG, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2609.35715v1 Reinforcement learning (RL) in simulation can train dexterous manipulation policies without robot demonstrations, but training a single generalist policy with task-agnostic rewards faces a severe exploration problem: approaching, grasping, and reorienting diverse objects with many degrees of freedom is difficult to discover from scratch. Prior works make exploration tractable with high-quality robot demonstrations, per-task reward shaping, or by restricting policies to narrow modes of behavior. We propose X-Reset, a framework that instead resolves exploration with human hand-object demonstrations. Rather than imitating or tracking retargeted human motion, X-Reset kinematically retargets hand-object states to noisy robot states, filters out states that are unstable in simulation, and samples the remainder as resets during RL training with general-purpose object-centric rewards. The resulting policy depends only on object state and goal, with demonstrations entering training through the reset distribution. We show that X-Reset trains generalist policies on 20 objects across three embodiments---a 22-DoF hand on two different arms and a parallel-jaw gripper---and resolves the exploration challenges of RL from scratch. X-Reset scales with the number of training objects, generalizes to unseen objects, can learn from imperfect hand-pose estimates, and transfers behaviors zero-shot from sim-to-real. --- ### Lagrangian--Hamiltonian Flows for Video Prediction and Image Generation: A Symplectic Perspective Authors: Jiawei Hu Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35710v1 We introduce LHFM, a geometric framework for learning image dynamics. Drawing on structures central to classical mechanics, symplectic geometry, and geometric quantization, LHFM represents each image as an exact Lagrangian graph and models its evolution through image-dependent Hamiltonian flows, which yield a transport--source parameterization of image velocities. Our primary application is deterministic video prediction: LHFM-V is a recurrent model that advances frames by integrating predicted transport and source fields, and achieves the lowest reported FLOP count among the compared recurrent models with similar prediction accuracy. The image variant, LHFM-I, shows that the same construction is compatible with flow matching: in a matched experiment, it attains a lower FID than the flow-matching baseline. --- ### Mind the RefGAP: Correcting Reference Attention in Diffusion-Based Visual Editing Authors: Yanan Wang, Shengcai Liao, Guangyi Liu, Xiaodan Liang Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35708v1 Reference-guided diffusion editors struggle to faithfully reproduce user-provided references. We identify a potential bottleneck in diffusion editors: many methods provide limited reference-attention allocation. For example, in LoomVideo, edit-region queries assign less than 1% of their attention mass to the reference. We introduce RefGAP, a training-free correction that determines logit-offset magnitudes online at each layer from the reference-attention mass measured during the forward pass. Positive offsets to reference logits strengthen reference usage by edit-region queries, while negative offsets for keep-region queries limit reference-induced changes outside the edit. Two global coefficients control the correction; they are selected once on validation data from four development diffusion editors and held fixed. Across seven diffusion-based image/video editors, RefGAP improves identity fidelity in head swapping and face swapping. RefGAP achieves a fidelity-preservation trade-off comparable to separately tuned constant edit-side biases, without per-approach strength sweeps. Additional experiments on virtual try-on and background replacement evaluate transfer beyond identity editing. --- ### ScAn-Bench: Evaluating Scaling Analysis Methodology Authors: Artin Sermaxhaj, Nastaran Alipour, Donat Sinani, Johannes Hog, Neeratyoy Mallik, Jenia Jitsev, Danny Stoll Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35707v1 Recent progress in machine learning is driven by large-scale foundation models, where scaling laws and finding optimal scaling prescriptions for architecture, data, and hyperparameters are key in advancing the state-of-the-art. Therefore, it is surprising that no systematic study evaluates the methodology to obtain scaling laws and prescriptions across different model types. To shed light on this crucial blind spot and facilitate future research, we introduce the surrogate benchmarks ScAn-Bench-LLM and ScAn-Bench-VLM based on 4524 and 8024 checkpoints of language and vision-language model pipelines. On our benchmarks, we perform the first systematic evaluation of both data acquisition and extrapolation methodology for scaling analysis across different data modalities. --- ### Reinforcing Agentic Creativity in Scientific Ideation with Night Science Authors: Priyanka Kargupta, Silviu Cucerzan, Shweti Mahajan, Allen Herring, Jiawei Han, Ryen W. White, Sujay Kumar Jauhar Published: 2026-09-28 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.35706v1 Large language models (LLMs) excel at structured, verifiable tasks, but their low-entropy bias can produce homogeneous and predictable outputs, limiting their utility for open-ended scientific ideation. Effective discovery, however, spans a broader creative spectrum: from structured day science to loosely structured, serendipitous night science that reaches ideas beyond those typically considered. We introduce AI Night-Scientist, an agentic framework that uses reinforcement learning to teach models when and how to depart from predictable reasoning. Grounded in cognitive science, we model creativity along three axes: action (what to do and how creatively), process (when to explore versus exploit), and outcome (the novelty and usefulness of the resulting idea). We use these axes to train models with GRPO, exposing them to varying degrees and forms of creativity throughout training. This produces substantially more diverse scientific proposals, expanding the range of research directions by 27.8% and contribution types by 14.9% over the base model. It also improves predicted citation impact by up to 32.0 percentage points and originality by 66.2 points. These gains cannot be reproduced by simply increasing decoding temperature; instead, we find that semantic guidance specifying what kind of creativity to pursue is critical. Overall, our results suggest that creativity is a learnable, multi-level ability that can be shaped to help researchers reach ideas beyond those typically explored by LLMs. --- ### DynaTokens: Teaching Dynamics to Camera-Controlled Video Models at Test Time Authors: Ziqi Ma, Hongqiao Chen, Georgia Gkioxari Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35704v1 Video generation must account for two sources of motion, one induced by the observer's camera path and the other caused by scene dynamics. An ideal camera-controlled video model should account for both motions: let users move the camera while evolving the scene dynamics. While current models handle camera-induced motion well in static settings, they struggle for dynamic scenes: objects are static, move incorrectly, or degrade in generation quality. We introduce DynaTokens, a lightweight set of learnable scene-specific tokens that teach dynamics to an existing camera-controlled world model. Our method is motivated by a simple asymmetry between the two sources of motion: whereas camera motion affects the generated view globally, object dynamics are spatially localized. Through cross-attention, DynaTokens trains the learnable tokens from a few example trajectories for a scene while keeping the base model frozen, and enables dynamics under new query camera paths. DynaTokens achieves a better simultaneous dynamics-camera tradeoff on VBench2 and WorldScore evaluations than LoRA, block finetuning, and specialized trainable-layer baselines. Analyses of token attention, ablations, and motion temporality suggest that matching the trainable interface to the structure of the learning target is important for effective adaptation. Project website: https://glab-caltech.github.io/dynatokens/ --- ### A Unified Uncertainty Representation for Graph Neural Networks via Doubly-Spectral Stochastic Expansion Authors: Fred Xu, Thomas Markovich, Florence Regol, Yizhou Sun Published: 2026-09-28 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.35703v1 Reliable deployment of graph neural networks requires calibration, out-of-distribution (OOD) detection, and robustness to distribution shift, yet existing methods address these needs with separate models and objectives. We model uncertain node embeddings as random graph signals: graph Fourier filters capture structural variation, and a scalar orthogonal-polynomial chaos coordinate captures latent stochastic variation. The resulting doubly-spectral stochastic (DSS) expansion supplies task-matched readouts from one representation: the mean coefficient encodes class evidence for the energy-based OOD score, the higher-order coefficients encode structured logit variation, and quadrature averaging over the chaos coordinate defines the single predictive distribution used for prediction and calibration. A capacity theorem shows that, under a full-rank feature assumption, a restricted subfamily matches the chaos coefficients of any Gaussian-latent random graph signal, with exponentially decaying truncation error under a growth condition; the task-level claims are established empirically. DSS-GNN has two deployment modes: standalone, or as a residual branch beside a deterministic encoder (DSS-Hybrid). Standalone DSS-GNN achieves the lowest Brier score among the compared uncertainty-aware baselines on all 14 node classification benchmarks without post-hoc correction; DSS-Hybrid achieves the best AUROC on most node-OOD settings, competitive cross-graph OOD detection, and the strongest shifted accuracy on all 7 GOOD concept-shift benchmarks under standard empirical risk minimization (ERM). Cross-evaluating both modes on all three tasks shows that each remains effective on the other's tasks, with documented exceptions, and yields explicit deployment guidance. --- ### MeqMuon: Matrix-Equilibrating Muon for LLM Pretraining Authors: Chang-Wei Shi, Xu Wang, Wu-Jun Li Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35701v1 The success of large language models (LLMs) has been accompanied by continued growth in model size and pretraining costs. Muon offers high accuracy and training efficiency in LLM pretraining. Recent work introduces row-wise normalization into Muon to balance update magnitudes and improve pretraining performance. However, row-wise normalization alone cannot accommodate different imbalance patterns in update matrices. In this paper, we propose an improved Muon optimizer, called \underline{m}atrix-\underline{eq}uilibrating Muon~(MeqMuon), for LLM pretraining. MeqMuon balances both row and column magnitudes through normalization that can be automatically tailored to different imbalance patterns without manual intervention. Moreover, MeqMuon eliminates the need to store AdamW's second-moment estimates, reducing optimizer-state memory usage. Empirical results demonstrate that MeqMuon achieves better convergence performance than AdamW, Muon, and other baselines in LLM pretraining. --- ### Distillation Defenses Easily Break After Reinforcement Learning Authors: Shidan Javaheri, Alexander Panfilov, Oliver Britton, Yarin Gal, Yonatan Gideoni Published: 2026-09-28 Categories: cs.LG, cs.AI, cs.CR arXiv: https://arxiv.org/abs/2609.35699v1 Distillation attacks copy the reasoning capabilities of closed-source large language models, allowing bad actors to replicate state-of-the-art performance at low cost. Attackers systematically collect a large volume of frontier model reasoning traces and then train (i.e., "distill") their own models on these traces. Existing defenses against distillation attacks are typically evaluated immediately after distillation, implicitly assuming attackers do not train their models any further. In this paper, we argue that a more realistic threat model includes further training with reinforcement learning after distillation. A misspecified threat model can give a false sense of security -- some defenses that seem effective after distillation can be broken after subsequent reinforcement learning. Practically, reinforcement learning lowers the bar for a distillation attack to be effective. We show that simple attacks can steal reasoning capabilities from existing closed-source language models using data easily obtainable from current APIs, yielding reasoning improvements equivalent to more sophisticated attacks that extract the full hidden traces. Results indicate that any distillation defense that leaks sufficient information to reconstruct approximate reasoning traces is likely ineffective. We conclude by discussing broader implications and batch-level distillation defenses which could be more effective. --- ### Provable Benefits of Regularization: Fast Rates for Adversarial Imitation Learning Authors: Hanbin Zhou, Shangzhe Li, Alexander Braverman, Weitong Zhang Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35698v2 We study adversarial imitation learning (AIL), in which an agent learns to imitate expert demonstrations by optimizing a policy against an adversarial reward that distinguishes expert and learner behavior. Historically, reward regularization and entropy-based policy regularization are key components of empirically successful methods such as GAIL and LS-IQ, yet their finite-sample benefits remain underexplored. We establish fast rates for jointly regularized AIL in finite-horizon Markov decision processes with general function approximation. Our model-free algorithm, Dually Regularized AIL, combines KL policy regularization with a quadratic reward penalty weighted by expert and learner occupancies. With K online episodes and N expert trajectories, we prove a $\widetilde{O}\left(\frac{1}{K}+\frac{1}{N}\right)$ bound on the regularized imitation gap for fixed regularization parameters. Our analysis combines an online mirror descent construction for general convex reward classes to control estimation error from finite expert data and stochastic learner feedback, with a sharp analysis of optimistic KL-regularized policy learning. To the best of our knowledge, Dually Regularized AIL is the first algorithm to simultaneously achieve $\widetilde{O}\left(\frac{1}ε\right)$ sample complexity in both expert demonstrations and online interactions for this regularized AIL objective, even with stochastic experts. These results provide a rigorous characterization of the complementary statistical benefits of reward and policy regularization in AIL. --- ### Provable Benefits of Regularization: Fast Rates for Adversarial Imitation Learning Authors: Hanbin Zhou, Shangzhe Li, Alexander Braverman, Weitong Zhang Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35698v1 We study adversarial imitation learning (AIL), in which an agent learns to imitate expert demonstrations by optimizing a policy against an adversarial reward that distinguishes expert and learner behavior. Historically, reward regularization and entropy-based policy regularization are key components of empirically successful methods such as GAIL and LS-IQ, yet their finite-sample benefits remain underexplored. We establish fast rates for jointly regularized AIL in finite-horizon Markov decision processes with general function approximation. Our model-free algorithm, Dually Regularized AIL, combines KL policy regularization with a quadratic reward penalty weighted by expert and learner occupancies. With K online episodes and N expert trajectories, we prove a $\widetilde{O}\left(\frac{1}{K}+\frac{1}{N}\right)$ bound on the regularized imitation gap for fixed regularization parameters. Our analysis combines an online mirror descent construction for general convex reward classes to control estimation error from finite expert data and stochastic learner feedback, with a sharp analysis of optimistic KL-regularized policy learning. To the best of our knowledge, Dually Regularized AIL is the first algorithm to simultaneously achieve $\widetilde{O}\left(\frac{1}ε\right)$ sample complexity in both expert demonstrations and online interactions for this regularized AIL objective, even with stochastic experts. These results provide a rigorous characterization of the complementary statistical benefits of reward and policy regularization in AIL. --- ### Rethinking Personalized Generation: Test-Time Alignment via Factorized Ranking Models Authors: Qiyao Ma, Junshan Zhang, Zhe Zhao Published: 2026-09-28 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2609.35695v1 Aligning large language models (LLMs) to diverse user preferences is fundamentally hindered by standard alignment paradigms that optimize for monolithic users. In this work, empirical studies are first used to reveal the existence of a massive, untapped performance headroom for personalized generation through test-time alignment. We demonstrate that personalized generation is uniquely suited for test-time scaling methods like Best-of-N (BoN) because it can be viewed primarily as a candidate matching problem rather than a generator capability bottleneck. While reward models could in principle exploit this headroom, they are poorly calibrated for personalization, and their billion-parameter scale makes scoring large candidate pools prohibitively expensive. To overcome this limitation, we propose a parameter-efficient framework utilizing million-parameter scale multi-layer perceptron (MLP) ranking models. Our personalized ranking model directly reuses the internal embeddings of the base generator with minimal overhead. By scaling train-time data to provide fine-grained personalized preferences, this million-parameter ranking model accurately scores large candidate pools and can seamlessly guide generation to reduce the cost of materializing N candidates. Extensive experiments on nine datasets spanning three personalized generation settings show that our personalized ranking model effectively exploits the discovered headroom, outperforming billion-parameter generalist reward models on every dataset, with under 0.4% of their parameters and four orders of magnitude lower scoring latency. --- ### Reasoning with Continuous Latent Diffusion Authors: Xiang Cheng Published: 2026-09-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.35694v2 Continuous diffusion generates complete reasoning solutions through iterative refinement in latent space. We introduce the Continuous Embedding Diffusion Reasoner (CEDR), an ELF-based training and inference recipe. Our experiments show that accurate decoding alone does not ensure strong reasoning performance. We therefore learn compact representations from multiple layers of a strong autoregressive teacher. Their decomposition also enables asynchronous denoising at different rates. We show that prompt encodings need only preserve the information required for the correct text-conditional score, rather than exactly match teacher features, and use a staged curriculum to learn a compact prompt encoder that replaces the teacher Transformer at inference. We adapt DiffusionNFT to learned self-conditioning guidance and incorporate gold-solution endpoints to supplement sparse rewards. Our supervised models outperform reported results from recent continuous-diffusion baselines at comparable backbone scales on mathematical reasoning and HumanEval code generation. With a 638M-parameter denoising backbone and learned prompt conditioning, post-NFT CEDR-L achieves 63.74% pass@1 on GSM8K and 24.6% on MATH500 at 64 denoising steps, and 32.85% on HumanEval and 30.18% on HumanEval+ at 128 denoising steps. Code will be available at: https://github.com/chengxiang/CEDR. --- ### Reasoning with Continuous Latent Diffusion Authors: Xiang Cheng Published: 2026-09-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.35694v1 Continuous diffusion generates complete reasoning solutions through iterative refinement in latent space. We introduce Latent Flow Reasoning Models (LFRMs), an ELF-based training and inference recipe. Our experiments show that accurate decoding alone does not ensure strong reasoning performance. We therefore learn compact representations from multiple layers of a strong autoregressive teacher. Their decomposition also enables asynchronous denoising at different rates. We show that prompt encodings need only preserve the information required for the correct text-conditional score, rather than exactly match teacher features, and use a staged curriculum to learn a compact prompt encoder that replaces the teacher Transformer at inference. We adapt DiffusionNFT to learned self-conditioning guidance and incorporate gold-solution endpoints to supplement sparse rewards. Our supervised models outperform reported results from recent continuous-diffusion baselines at comparable backbone scales on mathematical reasoning and HumanEval code generation. With a 638M-parameter denoising backbone and learned prompt conditioning, post-NFT LFRM-L achieves 63.74% pass@1 on GSM8K and 24.6% on MATH500 at 64 denoising steps, and 32.85% on HumanEval and 30.18% on HumanEval+ at 128 denoising steps. Code will be available at: https://github.com/chengxiang/LFRM --- ### Report: Progressive Disclosure of Agent Skills Authors: Guilin Zhang, Kai Zhao, Priyanka Mudgal, Waleed Ammar, Xiquan Cui, Xu Chu, Alet Blanken Published: 2026-09-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.35692v1 Users of Workday's deployed LLM-based agents often request features which can be addressed by defining named procedures, also known as skills, in the LLM context, effectively augmenting agents' capabilities. However, as an agent's skills library grows in size, so does the agent's operational cost. Progressive disclosure (lazy-loading) of skills as needed may reduce operational costs, but its impact on overall latency and skill-retrieval quality remains unclear. In this report, we investigate the impact empirically and find that progressive disclosure improves skill-retrieval quality but marginally degrades overall latency. --- ### Rethinking Circuit Evaluation: Do Circuits Explain Model Errors? Authors: Li Zhang, Chuqin Geng, Mark Zhang, Chen Yang, Luke Zhang, Haolin Ye, Xujie Si Published: 2026-09-28 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.35686v1 Mechanistic interpretability (MI) aims to explain a model's behaviour through analyzing its internal computations; circuit-based explanations aim to isolate these computations with compact subnetworks validated by ablating the rest of the model. We show that circuits validated this way may fail to recover the underlying mechanism of the model's behaviour by closely reproducing its successful decisions while failing to account for most of its errors. Such explanations should account for the model's particular errors as well as its successes. We evaluate this requirement by measuring exact answer agreement separately on model successes and failures, across circuit sizes and ablation settings, on IOI, Docstring, and six model-task settings from the Mechanistic Interpretability Benchmark. We discover that many tested circuits closely replicate correct behaviour while missing most of the model's errors. On indirect object identification (IOI) for GPT-2 small, under mean ablation, the manual circuit and tested automated circuits, including one trained against the model's full output distribution, agree with the model on 97.3-99.5% of prompts it answers correctly but only 11.4-41.7% of errors. An IOI case study shows that lost errors are recoverable by restoring omitted attention-heads which raise error reproduction from 14.2% to 75.1% on a separate held-out set with 0.41 percentage point decrease on correct agreement, exceeding matched random extensions and scalar-biased control. Intervention traces show how omitted computations produce specific wrong answers for a reproducible subset of errors. In all, these findings show circuits can preserve task success without adequately explaining model's failures, and support exact error reproduction as a necessary, but not sufficient, test of circuit-based explanations of model behaviour. --- ### QuanReview: Offline, Auditable Reconciliation of Human and LLM Span Annotations Authors: Matteo Musacchio, Juan Cruz Giner Pulero, Isabel Castañeda, Naomi Couriel, Yelena Mejova, Mariano G. Beiró, Kyriaki Kalimeri Published: 2026-09-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.35685v1 Structured span annotations, such as quantities with their units, uncertainty modifiers, and event classes, are expensive to create and hard to keep trustworthy once language models enter the loop. We present QuanReview, an open-source system for auditing and correcting such annotation layers. QuanReview aligns two annotation streams over the same documents at character level, resolves unambiguous cases by an explicit and logged policy, and routes candidate conflicts to a browser-based adjudication interface where reviewers accept either side, build field-level hybrids, or flag items for re-annotation. A campaign manager assigns documents to multiple annotators with configurable redundancy, computes agreement at document and span level, auto-merges unanimous documents, and exports the corrected layer in the original file format, so that it can replace the original annotation files directly. Applied to a 4,457-record humanitarian benchmark and an LLM extraction stream, the system fully auto-merged 8% of documents, applied automatic policy decisions to a further 1,513 records, and concentrated human attention on 3,131 candidate conflicts, a mean of 5.4 per reviewed document. --- ### The Hidden Perception Constraint in Task-Aware Compression Authors: Sahan Liyanaarachchi, Semih Akkoc, Sennur Ulukus, Aylin Yener Published: 2026-09-28 Categories: cs.IT, cs.LG arXiv: https://arxiv.org/abs/2609.35684v1 With the recent advancements of neural compressors, explicitly incorporating perception constraints into the design of compression schemes has gained significant attention. Traditionally, these perception constraints ensure that the distribution of the reconstruction does not significantly deviate from the distribution of the source, thus attesting to the perceptual quality of the reconstruction. In this work, we uncover several perception constraints that are naturally present in task-aware compression. In particular, we consider a problem where the primary task is reconstruction and the secondary task is classification (i.e., a statistical test). We study this problem at varying levels of domain information available to us and discuss how to utilize the naturally emerging perception constraints to design rate-minimal compression schemes that also maximize the utility of our secondary task. We show that in this setting, if the decision boundaries of the classifier are ill-defined (mismatch) for our source distribution, then matching onto a target distribution enhances our classification accuracy. --- ### Verifier Errors in RLVR: Reward Hacking, Limits of Feedback, and Selective Control Authors: Christian Moya, Elliott Thornley, Guang Lin Published: 2026-09-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.35677v1 In reinforcement learning with verifiable rewards (RLVR), imperfect verifiers can reward incorrect responses, creating opportunities for reward hacking. Using gradient flow with a fixed verifier, we characterize the conditions under which reward rises while correctness falls. We then show that the observations available during RLVR are, in general, insufficient to detect or identify accepted errors, or to guarantee their reduction without sacrificing correct responses. To address this limit, we construct a correction using additional feedback about correctness from audits. This correction achieves \emph{selective control}: at the current policy, it lowers the probability of accepted errors and raises that of correct responses, provided it outweighs the pressure toward errors from verifier reward. Experiments with log linear and neural contextual bandits and with a language model support the analysis and show that selective control under partial auditing reduces accepted errors while increasing correctness. --- ### Tracing the Evolution of Oracle Bone Characters Across Three Millennia Authors: Tianhao Fu, Xinxin Xu, Spike Wang, Cunyi Kang, Jian Cao, Xixin Cao Published: 2026-09-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.35674v2 Of the approximately 4,500 Oracle Bone Inscription (OBI) characters discovered from the Shang dynasty, only about 1,600 have been deciphered. Many computational approaches compare OBI with glyphs from one historical period at a time. However, during the evolution of Chinese characters, significant structural or semantic changes often occur in uncertain dynasties. A single-period reference may be insufficient when relevant forms change substantially between observed eras. Therefore, we propose the \textbf{Manifold-based Script Evolution Framework (MSEF)}, a framework that models the evolution series (OBI, Bronze, Seal, Clerical, Regular) of Chinese characters as the continual evolution of a manifold space. MSEF represents each character as an era-specific manifold point and learns continuous inter-era transition rules via Neural Ordinary Differential Equations. Both manifold space and transition dynamics can be trained end-to-end through character evolution pairs across any two eras. --- ### Tracing the Evolution of Oracle Bone Characters Across Three Millennia Authors: Tianhao Fu, Xinxin Xu, Spike Wang, Cunyi Kang, Jian Cao, Xixin Cao Published: 2026-09-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.35674v1 Of the approximately 4,500 Oracle Bone Inscription (OBI) characters discovered from the Shang dynasty, only about 1,600 have been deciphered. Many computational approaches compare OBI with glyphs from one historical period at a time. However, during the evolution of Chinese characters, significant structural or semantic changes often occur in uncertain dynasties. A single-period reference may be insufficient when relevant forms change substantially between observed eras. Therefore, we propose the \textbf{Manifold-based Script Evolution Framework (MSEF)}, a framework that models the evolution series (OBI, Bronze, Seal, Clerical, Regular) of Chinese characters as the continual evolution of a manifold space. MSEF represents each character as an era-specific manifold point and learns continuous inter-era transition rules via Neural Ordinary Differential Equations. Both manifold space and transition dynamics can be trained end-to-end through character evolution pairs across any two eras. --- ### FlowTool: Controlling Tool Parameter in Image Retouching via Flow Matching Authors: Thanh-Long V. Le, Steven Walton, Seunghyun Yoon, Branislav Kveton, Trung Bui, Eunho Yang, Viet Lai Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35673v1 Tool-based image editing (image retouching) is commonly formulated with autoregressive multimodal large language models (MLLMs) that sequentially generate reasoning, tool selections, and parameter values. In this work, we present a novel approach to tool-based image editing by framing the task as a flow matching problem. We introduce FlowTool, a framework that directly models the distribution of high-quality tool parameters conditioned on the input image and user instruction using conditional rectified flow. FlowTool combines a vision-language model backbone for multimodal understanding with a Diffusion Transformer parameter generator that transforms Gaussian noise into an editing plan. We train FlowTool with a two-stage supervised flow-matching curriculum, followed by reward-based post-training. Across MMArt-Bench, FlowTool-Eval, ArtEdit-Bench, and MIT-Adobe5K, FlowTool achieves significantly stronger reference-based performance than specialized MLLM editing agents and proprietary MLLMs, while remaining competitive with proprietary models under reference-free evaluation. Moreover, FlowTool significantly improves inference efficiency, reducing latency by at least $50\times$ while requiring nearly $2\times$ less memory than the compared baselines. These results demonstrate that tool-based image editing can be effectively modeled as conditional generation over structured continuous editing parameters, without autoregressive reasoning. --- ### PhoneCLI: From App Interfaces to Callable Commands for Mobile Agents Authors: Yangqin Jiang, Lingrui Xu, Chao Huang Published: 2026-09-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.35671v1 Mobile GUI agents operate through a perception--action loop: at each step they screenshot the device, invoke a vision--language model (VLM), and emit an action. It is slow, costly, and brittle, yet most of what it does is navigation---and everyday navigation is static, ordered, and endlessly repeated. We present PhoneCLI, which compiles an app's GUI navigation into callable commands, without any app-internal API, runtime instrumentation, or model training. Offline, PhoneCLI explores a target app from the outside and distills its screens, interactive elements, and navigation edges into a semantically annotated map; each screen yields one deterministic command: a replay sequence that reaches it. Online, the agent selects a command, verifies it before execution, and then executes it deterministically in sub-second time at zero VLM cost; open-ended interaction and every failure of the compiled path fall back to the embedded VLM interpreter, exactly the pure VLM agent, so compilation can only help. On AndroidLab, PhoneCLI improves the task success rate while reducing steps and token consumption, and it transfers to AndroidWorld's official M3A agent with consistent efficiency gains. What PhoneCLI compiles is the app's navigation rather than one run, so it serves new tasks, not only repeated ones. --- ### Learned Preconditioning for a Primal-Dual Interior-Point Method Authors: Abhinav Madabhushi, Jialin Liu, Minxin Zhang Published: 2026-09-28 Categories: math.OC, cs.LG arXiv: https://arxiv.org/abs/2609.35665v1 Interior-point methods (IPMs) are among the most widely used algorithms for constrained optimization, yet their Newton-based search directions require costly second-order information and large linear-system solves. Learning to optimize offers cheaper updates learned from data, but the singular behavior of logarithmic barriers near constraint boundaries makes IPMs highly sensitive to perturbations, complicating both warm starting and learning reliable updates. We introduce pdLIP, an IPM for smooth nonlinear programs that integrates learned preconditioning with pdProj, an all-shifted primal-dual projected-search IPM. A shared coordinate-wise recurrent network predicts a positive diagonal preconditioner that scales the right-hand side of the reduced Newton system for the primal step, and the remaining slack and multiplier directions are recovered analytically. The learned iterations avoid Hessian evaluations and Newton-system solves, using only first-order and coordinate-wise operations amenable to GPU parallelization. Training is self-supervised, with a loss based on a penalty-barrier merit function and the residual of perturbed optimality conditions, requiring neither target directions nor precomputed solutions. Primal and dual shifts mitigate the barrier's sensitivity to perturbations near constraint boundaries, enabling effective warm starting. Across four classes of 200-dimensional convex and nonconvex constrained problems, pdLIP warm starts reduce pdProj refinement iterations by 63-67% compared with cold starts at the same KKT residual tolerance of $10^{-8}$, with negligible warm-start generation cost relative to the subsequent pdProj solve. Improvements persist on box-constrained QPs with 1000 variables and extend to applications including portfolio optimization, support vector machines, and a nonlinear control example. --- ### MS-GLA: Multi-Scale Gated Linear Attention for Addressing Representational Bottlenecks via Multi-Temporal Resolution Authors: Prasoon Dev, Anirudh Sankar, Vasudeva Varma Published: 2026-09-28 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.35664v1 Gated Linear Attention (GLA) Transformers advance linear recurrent models through data-dependent gating, but face a core limitation: the fixed-capacity memory matrices across all heads operate at a single temporal resolution, where each token is processed individually, forcing them to simultaneously encode local syntactic patterns and long-range semantic structure, creating a representational bottleneck that gating alone is insufficient to resolve. We introduce Multi-Scale Gated Linear Attention (MS-GLA), which addresses this by distributing attention heads across multiple temporal resolutions. Coarser resolutions pool longer token spans naturally specializing toward long-range dependencies, while finer head groups retain sensitivity to local syntactic structure. A learnable, input-dependent fusion layer dynamically recombines head group outputs at each timestep, expanding effective memory capacity without increasing per-head state size. This multi-resolution decomposition draws on principles from Multi-Scale State-Space Models (MS-SSM), adapting them to the gated linear attention setting. We evaluate MS-GLA on language modeling, recall-intensive tasks, and long-context generalization. Across all settings, MS-GLA consistently achieves higher accuracy and lower perplexity than GLA at matched parameter counts, with up to 18.9% improvement on recall-intensive tasks and 9.5% lower average perplexity on language modeling benchmarks, validating multi-temporal resolution decomposition as a principled and effective extension of Gated Linear Attention. --- ### Late Attention Layers Alone Can Copy Entity Tokens, but Not Without Attending to Their Context Authors: Muyu He, Yuchen Liu, Ran Tao, Li Zhang Published: 2026-09-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.35663v1 Large language models (LLMs) reliably perform entity copying, in which a model copies tokens referring to an entity, termed entity tokens, from the prompt into its output to answer a question. Although entity copying is straightforward for most LLMs, existing research does not provide a systematic account of which layers specialize in this fundamental task or how other tokens in the same sequence, termed context tokens, influence the model's ability to copy the entity tokens. To address these questions, we conduct experiments on Qwen3-8B using two novel methods: genie-in-a-bottle, which controls exactly which layers can participate in an entity-copying task, and attention lobotomy, which cuts off specific tokens' attention to entity tokens without affecting the remaining attention distribution. We find that two distinct groups of layers in the second half of the model are both necessary and sufficient for entity copying. Moreover, in addition to the decoding position's attention to entity tokens, context tokens' attention to entity tokens also proves necessary for copying the exact tokens, even though context tokens do not store entity information themselves unless they satisfy particular semantic properties. Our findings establish the critical role of late layers in entity copying under the guidance of context tokens, calling for future work on how models propagate and consume entity information. --- ### Many Eyes, One World: Feed-Forward 3D Reconstruction from Mixed Cameras Authors: Qiaoge Li, Yifan Zhan, Haijun Yang, Haiyang Liu, Yiyi Cai, Chenchi Luo Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35658v1 Real-world capture is heterogeneous: perspective, fisheye, and $360^\circ$ panoramic images can coexist within a single reconstruction task, yet most feed-forward 3D reconstruction models assume perspective imagery and a uniform input representation. Recent models handling several camera types are either informed of the camera type for each view or reconstruct one image pair at a time. No single-pass method reconstructs mixed-camera tuples containing full panoramas from images alone. We present MEOW, a feed-forward system that jointly reconstructs metric pointmaps and camera poses from one N-view tuple mixing perspective, fisheye and full-panorama images, in a single forward pass from images alone: no calibration, distortion parameters, camera-type labels or poses are supplied for any view. Our guiding design philosophy is to treat heterogeneous-camera reconstruction as a data-adaptation problem rather than an architectural redesign. MEOW retains a perspective-pretrained backbone and learns heterogeneous cameras entirely from a procedural data engine, which renders each scene across a continuous manifold of camera models with exact rays and depth, and certifies covisibility for every camera-sampled training tuple. Trained on synthetic tuples only, MEOW transfers zero-shot to real captures: on heterogeneous 2D3DS tuples it achieves 79.9 mAA@30 against 53.8 for Wid3R given the camera type of every view; on our laser-scanned mixed-camera benchmark it registers every four-view mixed tuple with 79.4 AUC@30. The data engine, benchmark, and complete evaluation pipeline will be released. --- ### CMDO: A Cognitive Memory-Driven Optimization Algorithm for Adaptive Population-Based Search Authors: Mohammed Yusuf Mujawar, Shahram Rahimi, Noorbakhsh Amiri Golilarz Published: 2026-09-28 Categories: cs.NE, cs.AI arXiv: https://arxiv.org/abs/2609.35657v1 Population-based optimization methods often use previous search information through successful solutions, parameter adaptation, or operator performance, but they rarely retain the context in which a search behavior succeeded or failed. We introduce Cognitive Memory-Driven Optimization (CMDO), a derivative-free population-based optimizer that represents experience as the relationship between search context, search behavior, and observed outcome. CMDO organizes these experiences across working, episodic, and consolidated memory, retrieves them according to similarity with the current search state, and uses both positive and negative evidence to guide subsequent search. Retrieved experience does not replay previous candidate locations; instead, it selects search recipes that are reconstructed from the current population through exploratory, directed, and local search behaviors with adaptive search geometry. We evaluate CMDO on selected Blackbox Optimization Benchmarking test suite on COCO (BBOB/COCO) and Congress on Evolutionary Computation 2017 (CEC2017) problems against DE, CMA-ES, SHADE, GWO, HHO, and ORCA, and further study its application to seven-parameter photovoltaic model estimation using measured current--voltage data. The results show problem-dependent but competitive optimization performance, including the lowest median error among the compared methods on CEC2017 F10. More importantly, analysis of the search traces shows that context-dependent recall changes the distribution of executed search behaviors, while unsuccessful experiences remain available as negative evidence for later decisions, showing that accumulated experience directly influences subsequent search behavior. These results support the use of explicit context--behavior--outcome memory as an active mechanism for controlling population-based search. --- ### Guided Uncertainty-Aware Robust Domain Transfer Authors: Xin Xiong, Zijian Guo, Tianxi Cai Published: 2026-09-28 Categories: stat.ME, stat.ML arXiv: https://arxiv.org/abs/2609.35654v1 Artificial intelligence models deployed as clinical decision support tools often suffer substantial performance degradation over time and across institutions due to covariate shift, concept drift, and cross-system heterogeneity. Fully retraining complex models is frequently infeasible, particularly in EHR settings where labeled outcome data are scarce and regulatory constraints limit model modification. We propose GUARD (Guided and Uncertainty-Aware Robustness to Domain shift), a unified statistical framework for updating an existing deployed model using limited labeled target data and multiple heterogeneous source populations while guarding against future distributional shifts. GUARD formulates robust multi-source transfer learning as a semi-supervised, adversarial optimization problem anchored at the current model, and employs double machine learning with cross-fitting to obtain debiased, efficient estimates that leverage abundant unlabeled target covariates. It then constructs an uncertainty-aware, guided group DRO estimator that combines source and target information while accounting for sampling variability in the source models. Our framework jointly provides principled robustness to domain shift, efficient semi-supervised estimation, and valid statistical inference for multi-source transfer of clinical prediction models. We demonstrate the utility of GUARD via extensive simulation experiments and a real world application to predicting future disease activity in rheumatoid arthritis from EHR data spanning more than a decade. GUARD recalibrates a pretrained model with only a small number of labeled records per year and remains accurate over multi-year horizons where target-only and standard transfer estimators degrade sharply. --- ### Transferable Mass Spectrum Prediction via Reference-Guided Test-time Specialization Authors: Yunhua Zhong, Runting Li, Yifan Li, Pan Liu, Zhiwen Yang, Zikun Wang, Yixuan Tang, Jun Xia Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35649v1 Tandem mass spectrum prediction supports compound identification across metabolomics, natural-product discovery, and environmental analysis. However, pretrained predictors often degrade under shifts in chemical space and acquisition conditions, while retraining domain-specific models from scratch is costly. We introduce SPARC, a retrieval-guided test-time specialization framework that adapts a pretrained predictor using a spectral reference library without accessing test-query spectra. For each target query, SPARC retrieves chemically related reference spectra to recalibrate fragment intensities within the learned fragmentation space. During Transfer, SPARC combines reference-guided spectral adaptation with reliability-aware consistency, using reconstruction behavior on retrieved spectra to selectively preserve trustworthy predictions during continual specialization. Across MassSpecGym, NPLIB1 and application-specific GNPS libraries, SPARC improves spectral prediction under multiple transfer settings. These results establish retrieval-guided test-time specialization as a practical strategy for extending pretrained MS/MS predictors to specific chemical and acquisition domains, with continual test-time training providing further refinement during deployment. --- ### Rubric Rewards from Item Response Theory Authors: Milad Yazdani, Yaser Souri, Xiren Zhou, Pranit Chawla, Dena Shahriari, Subhojit Som, Xia Song Published: 2026-09-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.35646v1 Many language tasks have no single answer that can be checked automatically. Rubrics provide criteria for judging responses to these tasks. For reinforcement learning, the resulting verdicts must be combined into a scalar reward. A common approach sums the points assigned to satisfied criteria. Distinct verdict patterns can thus receive the same reward, and the fixed points encode how much each criterion should count, not how strongly its verdict distinguishes the current rollouts. Beyond this aggregation problem, judging the full rubric needs more judge requests as the criterion count grows. To address these limitations, Rubric Response Theory (RRT) measures quality and selects criteria when rubric criteria are monotone indicators of a shared target. Rather than adding assigned points, RRT uses a two parameter item response model that treats the verdict pattern as evidence about scalar quality specific to the rubric. Under this model, its likelihood score maximizes the local signal-to-noise ratio for quality. Its Response Parameter Network (RPN) reads the prompt and criterion text to predict criterion difficulty and discrimination. As the policy distribution changes during training, RRT uses online expectation maximization to update the RPN from current rollout verdicts. With Qwen3.5-4B as the policy, RRT's macro criterion score across Medical, Science, Rubrics as Rewards Science, and RubricBench is 1.7 points above that of group relative policy optimization (GRPO). On hard and very hard criteria in Medical and Science, RRT gains 2.8 to 5.6 points over GRPO. At half the criterion budget, adaptive Fisher selection with a frozen RPN keeps the macro criterion score across four datasets within 0.1 points of GRPO with full judging. These results show RRT can reduce judge requests while remaining competitive with GRPO. --- ### CoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings Authors: Shama Gupta, Hoang H Nguyen, Chelsea Huang, Lindsay Devon Brin, Fanny Riols Published: 2026-09-28 Categories: cs.SD, cs.CL, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2609.35645v1 Code-switching (CS), a seamless alternation between languages within a single utterance, remains a critical challenge in automatic speech recognition (ASR). While prior works focus on conversational CS-ASR, enterprise settings demand evaluation of operational impact beyond edit-distance errors: how code-switching transcription errors propagate to downstream voice agent task failures. In this work, we propose (1) a CS-ASR synthetic benchmark and multidimensional evaluation framework tailored to enterprise domains, (2) systematic evaluation of frontier ASR systems across 5 language pairs, (3) diagnostic analysis of the additional transcription errors that code-switching introduces across language pairs and models. We release COSE-E to support enterprise-focused CSASR evaluation for multilingual voice agents in enterprise deployment. --- ### Not All Thinking is Created Equal: Latent Reasoning Discovers a Recurrent Search Algorithm for Depth Generalization Authors: Huzi Cheng, Zhewei Zhang Published: 2026-09-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.35643v1 Large Language Models can perform multi-step reasoning and improve task performance through different forms of intermediate computation, from token-based traces to computation carried out in latent space. However, a question remains open: do these different forms of thinking rely on the same underlying mechanism? To address this, we train and compare five variants of the same GPTNeoX backbone from scratch on an extended multi-hop reasoning task (ProsQA-Ext): a vanilla model, a Chain-of-Thought (CoT) model, a Pause Token model, and two latent-reasoning models that are optimized end-to-end without intermediate reasoning traces. We find that, strong in-distribution (ID) performance does not guarantee depth generalization. Vanilla, CoT, and Pause Token models solve ID problems well, but rely largely on local graph features and generalize poorly to out-of-distribution (OOD) problems with longer hops. In contrast, latent variants generalize better and show internal dynamics consistent with forward reachability propagation on the graph. Causal interventions and circuit analysis localize this computation to a sparse recurrent search circuit in the bottleneck latent model: an attention head retrieves graph relations, an MLP and the residual stream update the reachability state across recurrent steps, while multiple attention heads together then do the candidate matching. Together, these results show that different thinking mechanisms can learn distinct computational solutions, even at similar ID performance. In this setting, latent recurrence supports a reusable forward-search algorithm that generalizes beyond the training depth. --- ### Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts Authors: Shuyue Stella Li, Xiaochuang Han, Yulia Tsvetkov, Luke Zettlemoyer Published: 2026-09-28 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2609.35641v1 Precise instruction following in image generation, such as satisfying object counts and spatial relations, remains an open challenge at least in part because it is learned using unreliable reward models such as object detectors and vision-language models. We introduce Verifiable Visual Rewards (VVR), the first framework for programmatically verifiable image rewards, and show that training on it generalizes to natural prompts. Each VVR task is a scene of geometric objects and relations among them, from which we derive both the prompt and a deterministic verifier, so tasks can be generated in any number and at any chosen complexity. We release VVRBench, with 10,000 tasks over 32 constraint types, and VVRBench-Challenge, with 720 more complex tasks; the strongest model we evaluate---GPT-Image-2.5---solves 21.4% of VVRBench-Challenge. Using VVR scores as rewards for reinforcement learning (RLVVR) raises the accuracy of Stable Diffusion 3.5 Medium on VVRBench from 2.8% to 28.3% and demonstrates consistent easy-to-hard generalization. These gains extend to out-of-domain benchmarks, and mixing VVR into existing objectives further improves overall performance and human preference, motivating the adoption of VVR into standard image generation post-training recipes. --- ### GPUPhysBench: Benchmarking Coding Agents for Correct and Efficient GPU Physics Simulation Authors: Yuchen Sun, Jinjin He, Sinan Wang, Bo Zhu Published: 2026-09-28 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2609.35639v1 Writing fast GPU code for physical simulation is difficult: implementations must preserve numerical accuracy while handling irregular data access, synchronization, and iterative solvers. We introduce GPUPhysBench, a benchmark of 50 tasks testing whether coding agents can meet these demands. Tasks cover fluids, deformable solids, and granular materials, from individual simulation operators to complete simulators. Agents write, compile, test, and optimize GPU code with access to a NVIDIA GPU under fixed time budgets. We report pass rates and runtime performance relative to expert-optimized reference implementations. In a single-attempt evaluation of six frontier model-harness pairs, the two strongest pass all 50 tasks, but even the fastest reaches at least 0.9 the reference speed on only 22% of them, and no submission is more than 5% faster than the reference. The largest gaps arise in collision detection, constraint solving, and iterative solvers. GPUPhysBench brings physical simulation workloads to coding-agent evaluation, testing both the ability to implement numerical methods correctly and the ability to make them run efficiently. --- ### RT-Super: Learning Tumor Segmentation from Longitudinal Images and Reports Authors: Pedro R. A. S. Bassi, Wenxuan Li, Hanxue Gu, Jieneng Chen, Xinze Zhou, Zheren Zhu, Sezgin Er, Ibrahim E. Hamamci, Bjoern H. Menze, Gulhan E. Akan, Kang Wang, Yang Yang et al. Published: 2026-09-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.35637v1 Multi-tumor segmentation is important for early cancer detection and allows radiologists to visualize, verify, and understand AI predictions. However, tumor segmentation masks are expensive, time-consuming, and unavailable for many tumor types in public data. Instead, hospitals have vast, readily available data that can guide segmentation: radiology reports, longitudinal images, and multi-phase images. We use this readily available data to substitute for tumor masks in training AI for tumor segmentation. To this end, we propose a new architecture, RT-Super. It has a teacher network, which analyzes the patient's longitudinal images and reports to create high-quality tumor masks. These masks train a student network, which sees a single image and no report. At inference, when longitudinal images and reports are unavailable, we use the student. RT-Super uses a new CNN-Transformer architecture and novel Consistency Losses that exploit tumor location consistency across longitudinal images. We train RT-Super to segment esophagus, uterus and spleen tumors, which have few or no public masks. Even without training masks, RT-Super can segment these tumors and surpass public AI models. Overall, we demonstrate that learning from longitudinal images, multi-phase images, and reports can overcome mask scarcity and advance multi-cancer detection and segmentation. Code: https://github.com/MrGiovanni/RT-Super --- ### Bounding Retraining Equivalence and the Deletion Floor in Materials Machine Unlearning Authors: Can Polat, Mustafa Kurban, Erchin Serpedin, Hasan Kurban Published: 2026-09-28 Categories: cs.LG, cond-mat.mtrl-sci, quant-ph arXiv: https://arxiv.org/abs/2609.35635v1 In materials machine learning, closely related retained structures can sustain accurate property predictions even after removing a specific record, rendering post-deletion prediction error an ambiguous metric for machine unlearning. To resolve this ambiguity, we define the deletion floor as the expected target loss under a specified retraining procedure at the deleted request. Standard indistinguishability constraints yield a sharp interval bounding an update's target loss around this baseline reference. Theoretically, a conditional neighbor bound links a low deletion floor directly to retained fit, prediction regularity, and local label agreement, while an exact ridge identity isolates residual fit from the prediction change induced by record deletion. Empirically, controlled redundancy sweeps show an $\approx 8\times$ drop in median normalized retraining loss when one retained relative remains after deletion. Across two distinct fitting regimes in a paired Materials Project study, the lower-floor regime also exhibits a larger prediction change on more than 50% of the shared requests. Systematic comparisons against approximate updates and the original model decouple deliberate target suppression from preserved overall model utility. Consequently, request-level unlearning evaluations should report reference loss, prediction change, and retained utility together, interpreting post-deletion accuracy against what retraining itself leaves behind. --- ### DR-net-Mamba: Selective State-Space Modeling for Long-Range ECG Time-Series Denoising Authors: Basile Morel, Samuel Ruiperez-Campillo, Andreas P. Streich, Julia E. Vogt, Thomas Hofmann Published: 2026-09-28 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.35634v1 Electrocardiogram (ECG) recordings are corrupted by non-stationary noise sources that degrade diagnostic reliability, particularly in ambulatory and long-duration recordings. Deep learning denoisers exist, but convolutional architectures are limited by their receptive field, transformer-based models scale quadratically with sequence length, and diffusion-based approaches incur prohibitive inference cost. We propose a Mamba-augmented model that inserts selective state-space blocks at the convolutional bottleneck, combining local feature extraction with long-range temporal modeling at linear complexity. We comprehensively evaluate the proposed model with respect to reconstruction fidelity, noise robustness, recording-length scaling, and downstream diagnostic classification across over 40 pathology classes. On synthetic and real datasets, our model achieves the highest SNR and lowest RMSE, with the Mamba advantage increasing with sequence length and in low-SNR regimes. On classification with two independent classifiers, the proposed Mamba-based models achieve the best macro AUROC among all denoisers and improve over their convolutional base models. Calibration is more nuanced and classifier-dependent: denoising improves Binary Cross-Entropy and Brier score on Inception1D but often fails to beat the noisy input on ResNet1D-Wang, and the lead-specific Mamba variant is the only denoiser to improve both calibration metrics over the noisy baseline on both classifiers. Per-class analysis reveals a morphology-dependent benefit: Mamba substantially improves ST/T-change diagnoses, which depend on broad, context-sensitive waveforms. --- ### Which the Eye Fears: Writing with Read-Blindness Explains Massive Activations in Transformers Authors: Swagatam Mukhopadhyay, Vishal Vivek Saley, Vraj Parikh, Mausam Published: 2026-09-28 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.35630v1 Massive activation features (MAs) in Transformers are extreme-value residual-stream features that persist across layers despite the model's ability to suppress them. Why do they survive? Our investigation using an operator-level mechanistic analysis of attention and feed-forward (FFN) blocks reveals that these blocks systematically ignore MA coordinates while reading, but not while writing; creating a read-write asymmetry that blocks corrective feedback while allowing continued accumulation. We find that both attention and feed-forward layers have this read-blindness, and contribute to the emergence and persistence of MAs. To validate prior work that hypothesized that FFN's amplification abilities is the primary reason for MAs (Sun et al., 2026), we analyze the model checkpoints during learning. Contrary to our expectation, read-blindness emerges before FFN amplification, suggesting that it acts upstream in the MA mechanism. We further contribute gradient analysis to link this behavior to surprising asymmetries in the loss landscape, concluding that the model actively maintains this read-blindness. Finally, we find that removing read-blocking at different locations induces compensatory shifts elsewhere, but MAs still persist. --- ### SANTA++: Sampling Attention through Representative Keys Authors: Kyle Lee, Christian Z. Pratt, Ruoyu Fang, Heekyung Lee, Avinash Lohitsa, Ryan Modafe, Kerem Y. Camsari Published: 2026-09-28 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2609.35629v1 Attention often concentrates on a small subset of tokens in the context, but which subset matters changes from one query to the next. To exploit this changing structure, we introduce SANTA++, a training-free stochastic attention method that uses representative keys for memory-efficient selection without scanning the entire key-value (KV) cache. Cached keys are organized into teams, and the query scores one representative from each team to decide which teams to sample. We compute exact attention scores within the sampled teams and reweight each team's contribution by the inverse of its inclusion probability. This importance sampling correction estimates attention over the full cache, with a sampling budget that lets us trade memory reads for accuracy. Remarkably, with 32 or 64 sampled teams, SANTA++ uses 16% to 22% of dense attention's KV reads and retains 94% to 99% of the dense-attention baseline's scores on LongBench v2 and HELMET's retrieval-augmented generation subset, and 85% to 91% on RULER, with Qwen2.5-7B-Instruct at 32K context. With 31 sampled teams, our GPU implementation delivers a $1.69\times$ attention speedup over the dense FlashAttention baseline at 32K context. By reducing the number of cache entries read, SANTA++ in principle complements architectures with compressed KV representations, such as multi-head latent attention. Our kernels are available at: https://github.com/OPUSLab/santapp-kernel-demo.git. --- ### Arbitrary-Accuracy Neural Approximation with Optimal Neuron Count and Near-Optimal Bit Complexity Authors: Zilan Cheng, Li-Lian Wang, Zhongjian Wang Published: 2026-09-28 Categories: cs.LG, cs.NE, stat.ML arXiv: https://arxiv.org/abs/2609.35628v1 We study the minimum number of hidden neurons required for arbitrary-accuracy approximation of multivariate Hölder-continuous functions on $(0,1)^d$ and the associated encoding complexity. For $d\geq 2$, we construct a fixed, explicitly defined activation function for which a closed-form network with two hidden layers of widths $d$ and $1$ achieves arbitrary accuracy in the uniform norm. We prove that $d+1$ is the exact minimum total number of hidden neurons among standard feedforward networks with locally integrable activations and affine outputs. We further give a simpler construction using a single elementary activation that combines the floor and exponential functions. This construction requires three hidden layers of widths $d$, $1$, and $2$, only two neurons above the minimum. If a skip connection is allowed, widths $d$, $1$, and $1$ suffice. These constructions use explicit grid addressing and integer encoding of quantized function values. For a bounded $α$-Hölder class, they require $O(\varepsilon^{-d/α}\log(1/\varepsilon))$ bits, matching the metric-entropy lower bound up to a logarithmic factor. --- ### Can LLMs Value the Right Evidence? Evidence-Value Misalignment in Dynamic Medical Diagnosis Authors: Kehua Feng, Yunsheng Lu, Yitong Qiao, Tiantian He, Lei Liu, Yue Shen, Jian Wang, Jinjie Gu Published: 2026-09-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.35627v1 A correct diagnosis reached from insufficient or misleading evidence can pose a clinical hazard, yet outcome-based accuracy may reward such lucky guesses. We call this mismatch between diagnostic decisions and the value of available evidence Evidence-Value Misalignment (EVM). To disentangle evidential grounding independently from diagnostic accuracy, we introduce MedEVM, a dynamic benchmarking environment comprising 1,050 cases across 24 disease systems. Observations arrive turn by turn, requiring models to continuously calibrate its decision by deciding whether to wait for more evidence or submit a diagnosis. Across 9 LLMs, four interesting patterns are observed. (1) Miscalibrated evidence tracking. Making a diagnosis often fails to calibrate evidence sufficiency, even in more capable models, and even worsens in reasoning mode. (2) Misaligned diagnosis submission. Confidence in the correct diagnosis often fails to ensure timely submission despite sufficient evidence. (3) Evidence order matters. Reordering the same evidence changes diagnoses even when model confidence remains similar. (4) Misleading evidence remains influential. Added misleading evidence redirects diagnoses even after prior evidence becomes sufficient. We further verify that EVM predicts errors and that preventing premature submission improves accuracy. These findings motivate Evidence-Verified Diagnosis Harness (EVD-Harness). It decouples diagnosis generation from submission through an offline Contrastive Diagnostic Wiki and three online control stages, namely observation management, proposal and witness verification, and diagnosis submission control. Across five LLMs, EVD-Harness improves accuracy by 12.0--51.1 percentage points while mitigating EVM-related failures. Our results demonstrate that verifying evidential support before submission can make diagnostic decisions more reliable. --- ### RIDE: Reference-Anchored Inference-Time Diffusion Editing for Scaffold Hopping Authors: Ruoxi Gao, Frazier N. Baker, Trieu Nguyen, Xia Ning Published: 2026-09-28 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.35623v1 Scaffold hopping is a critical task in drug discovery, which seeks to discover new, structurally distinct molecules that share key functional groups and similar 3D shape with a reference binding ligand. Existing diffusion-based scaffold hopping methods formulate the problem as conditional generation of scaffolds given the functional groups. However, they lack a principled mechanism to jointly enforce 2D structural novelty and preserve the 3D shape of the reference ligand. Here, we introduce RIDE, a Reference-anchored Inference-time Diffusion Editing framework for scaffold hopping. RIDE recovers the reference diffusion noise trajectory conditioned on the binding pocket and functional groups, selects an optimal trajectory segment for editing via noise perturbation, and conducts a value-guided scaffold sampling to generate new scaffolds. Extensive experimental results demonstrate that, compared to baselines, RIDE consistently generates scaffolds with lower 2D similarity and higher 3D similarity to the reference, with an average improvements of 11.7% and 7.3%, respectively. Further analysis reveals that RIDE can accommodate various reward functions, and can preserve 3D similarity even when this is not explicitly included in the reward. Two case studies illustrate RIDE's ability to generate distinct scaffolds with different structures and properties, and its ability to introduce substantial 2D variation while maintaining very high 3D similarity. RIDE is publicly available at https://anonymous.4open.science/r/RIDE-C8A0. --- ### Elicitation and Decision Geometry in Single-Index Bandits Authors: Sakshi Arya, Cheng Soon Ong Published: 2026-09-28 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2609.35622v1 We study two-arm contextual bandits with arm-specific single indices and a shared unknown monotone link. Monotonicity makes the optimal action depend only on the contrast between the index directions, hence arm-specific reward functions need not be estimated. We introduce Natural Boundary Learning (NBL), a greedy procedure that uses a sequential Stein contrast to learn the optimal boundary directly, without estimating the reward functions or the common link. We characterize the local Riemannian dynamics of NBL through a decision stability coefficient balancing arm separation, link geometry, and the context distribution. We show that this stability is connected to the elicitation geometry of the underlying convex potential. Under local decision stability, NBL contracts toward the optimal boundary and achieves $O(\log n)$ expected regret. Numerical experiments illustrate the predicted stability regimes and compare NBL with a parametric greedy benchmark under link misspecification. --- ### Cartridges++: KV Cache Compression without Off-Context Derailment Authors: Sonia Laguna, Joao Monteiro, Marco Cuturi, Pierre Ablin, Eleonora Gualdoni Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35621v1 Serving long documents to a Large Language Model (LLM) repeatedly is expensive: computations grow with context length, and the memory footprint of the key-value (KV) cache balloons. Compressed KV (CKV) representations aim to mimic the cache of a document and are typically computed once and for all, ahead of inference time. Methods to obtain CKVs range from drop mechanisms that reduce their number of columns, to learned approaches. Among the latter, Cartridges have emerged as a leading compression method, learning compact KV representations through distillation on relevant Q/A pairs. While existing evaluations focus primarily on whether Cartridges and other CKVs yield approximately similar responses to document-related, on-context queries, we investigate the crucial deployment question of whether they can handle off-context queries, something the native KV representation is particularly good at, thanks to the mechanics of attention. We observe a fundamental trade-off: while Cartridges perform better for on-context queries, heuristic-variants preserve better the original LLM's ability to operate off-context. We measure this through their capability to avoid context contamination in their response, retain general knowledge, and follow instructions. We propose Cartridges++, simple modifications to cartridges that retain off-context abilities at small or negligible cost. The router variant decides at inference time whether the query should use the learned long-context memory, while the data-mixing variant allocates a small fraction of training Q/As to queries outside the reference long document. Our study shows that assessing CKVs on document utility alone can mask substantial degradation in broader model capabilities, yet those issues can be fixed with benign changes to CKV inference or training. --- ### Attention Graphons: A Graph Limit Perspective on Graph Transformers Authors: Caio F. Deberaldini Netto, Moshe Eliasof, Luana Ruiz Published: 2026-09-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.35620v1 Graph Transformers produce, for each attention head, a dense $n\times n$ matrix of learned pairwise interactions. We ask a fundamental question: do these attention-induced graphs converge to a stable limit object as $n$ grows, or does the learned interaction pattern remain unstructured and size-dependent? We answer this using dense graph limit theory, treating each attention matrix as a finite sample from an underlying kernel---an \emph{attention graphon}---and studying concentration around this limit under the cut-distance. We derive a worst-case variance bound requiring no assumptions on the graphon, and a sharper regularity-aware bound based on nonparametric estimation theory. To operationalize the theory, we propose a canonicalize-then-block-average pipeline for estimating dataset-level attention graphons, and a variance-based diagnostic for testing whether attention admits a stable continuum description. Experiments across multiple graph benchmarks show that learned attention stabilizes to dataset-specific graphon structure on several datasets; that empirical cut-distance and cut-norm variance decreases with $n$ consistent with our bounds; and that attention graphons transfer to larger graph sizes with error decreasing in $n$. --- ### From cacophony to hierarchy: a principled framework for assessing AI consciousness Authors: Shamil Chandaria, Arvo Muñoz Morán, Fernando Rosas, Anil Seth, Henry Shevlin, Marcus Hutter, Thore Graepel, Adam Bales, Iulia Comsa, Murray Shanahan, Ruben Laukkonen, Morten Kringelbach et al. Published: 2026-09-28 Categories: cs.AI, cs.CY arXiv: https://arxiv.org/abs/2609.35618v1 The question of AI consciousness is one of the most urgent pre-emptive problems in philosophy and computer science, yet progress is hampered by a cacophony of competing theories that often talk past each other. Separating the hard problem from the mapping problem allows the deepest metaphysical disagreements to be set aside: granting that experience supervenes on a system's organisation, the tractable question becomes at which grain of description that supervenience base sits. We extend Marr's three levels of analysis into a five-level hierarchy of functional descriptions (behavioural, computational, intrinsic causal-structural, organismic, and organism-environment) grounded in supervenience, coarse-graining, and multiple realisability. The major theories of consciousness are positioned within this hierarchy according to which level they take to be critical, and for each level we develop operationalisable indicators and assess current AI systems against them. A Bayesian model then combines theoretical credences with indicator evidence into an overall credence in a system's capacity for consciousness. In illustrative assessments, the verdict for current LLMs is driven as much by where theoretical credence is placed as by how the evidence is read: under different stipulated readings and credence distributions, assessments range from below 0.01 to roughly 0.8, showing sensitivity to assumptions. Finally, the consciousness indicators at each level closely overlap with the architectural features needed for general intelligence, suggesting that increasingly capable AI may become a stronger candidate for consciousness. The framework supports a structured agnosticism, in which theoretical commitments are made explicit, credences are updated as evidence accumulates, and assessments take the form of aggregated probabilities rather than verdicts. ---