# Research Papers — 2026-09 · bundle page 96 of 266 > Full-text markdown of papers #4751–4800 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/96 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/95.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/97.md) ### Even Sharper Bounds for Transductive Learning and Its Applications Authors: Yingzhen Yang Published: 2026-09-23 Categories: cs.LG, cs.IT, math.ST arXiv: https://arxiv.org/abs/2609.28459v1 We introduce Sharper Transductive Local Complexity (STLC), a localized complexity method for transductive learning under uniform sampling without replacement. The construction starts from a Bernstein-type concentration inequality for the supremum of the test--train empirical process. Its proof uses the modified log-Sobolev inequality for the swap walk and a two-parameter entropy closure. A peeling argument with a surrogate localization functional then gives excess-risk bounds with the same fixed-point and confidence terms as the classical inductive local Rademacher-complexity bounds, without the additional logarithmic confidence factor in earlier transductive results. For realizable learning over a binary class of VC dimension $\dVC$, with training size $m$, test size $u$, and $u\ge m\ge\dVC$, STLC yields $\cO\{\dVC\log(me/\dVC)/m\}$. This matches the standard inductive rate and, when $m\ge9$, is within a logarithmic factor of the transductive minimax lower bound of order $\dVC/m$. For transductive kernel learning, STLC gives a spectrum-adaptive excess-risk bound without the multiplicative imbalance factors appearing in the earlier local-complexity bound. --- ### Can LLMs Reason About Runtime Behavior? A Repository-Level Dynamic Benchmark Authors: Hamed Taherkhani, Mohammad Abdollahi, Melika Sepidband, Hridya Dhulipala, Tien N. Nguyen, Hadi Hemmati Published: 2026-09-23 Categories: cs.SE, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.28449v1 Large language models (LLMs) are increasingly used in coding tasks, but their ability to reason about code execution remains unclear. Existing repository-level QA benchmarks mainly evaluate static code understanding and often rely on LLM-based evaluation, while execution-reasoning benchmarks are mostly limited to snippets or functions. We introduce SWE-Flux, a repository-level benchmark for dynamic execution reasoning containing 480 execution-grounded instances across 12 real Python repositories, with gold answers automatically harvested from instrumented test executions rather than written manually or judged by LLMs. The benchmark covers singletest and multi-test questions over control flow, loops, program state, dataflow, exceptions, and program invariants. Evaluating five LLMs shows that this task remains challenging. The best model achieves only 37% accuracy. Models perform better on localized behavior such as invariants, intra-procedural control flow, exceptions, and simple loops, but struggle with dataflow, inter-procedural execution, precise state reasoning, and suite-level aggregation. Finally, we show that the oracle-harvesting pipeline can generate fresh benchmark variants using input perturbation. It successfully harvests valid variants for almost 90% of the selected instances, and the resulting variants are substantially more challenging for the evaluated models. --- ### Nonequilibrium Phases of Repulsive Self-Attention: Chaos, Attention Condensation, and Emergent Locality Authors: Qucheng Gao, Zuyi Yang, Xiao Chen Published: 2026-09-23 Categories: cond-mat.dis-nn, cond-mat.stat-mech, cs.LG arXiv: https://arxiv.org/abs/2609.28448v1 We study the nonequilibrium dynamics of a minimal recurrent transformer with $N$ normalized tokens, $Q=K=I$, and a negative value map $V=-I$. Similarity-based attention selects nearby representations, while the negative value map drives tokens away from the selected field. This feedback can continually reorganize both the representation geometry and the attention network. For $d=2$, the tokens lie on a circle, where the regular polygon is an exact fixed point. As the attention feedback strength $γ$ is increased, the polygon loses stability through a flip bifurcation, giving rise to period-two motion, chaos, and cluster-exchange or cluster-flip states. Despite this temporal complexity, attention remains diffuse as $N\to\infty$ at finite fixed softmax sharpness $β$. Attention condensation instead emerges in the scaling regime $β\sim N^2$. In the hard-routing limit, repulsive updates amplify local perturbations and routing-partner switches transmit them ballistically, producing an emergent butterfly cone in representation space. High-dimensional geometry provides a distinct route to localization. For $d=N\to\infty$, simulations from Gaussian initial conditions provide evidence for a condensation transition at $β=O(1)$, driven by dynamically generated finite overlap gaps. Depending on $γ$, the resulting phases include diffuse simplex-like states, consensus flips, condensed active routing with signatures of chaos, and fragmented cluster flips. These results establish temporal activity, attention condensation, and geometric clustering as distinct collective phenomena, and show that sparse attention can sustain persistent dynamics rather than freeze it. --- ### Order-Invariant Answers, Order-Sensitive Representations in Mathematical Reasoning Authors: Zhixu Silvia Tao Published: 2026-09-23 Categories: cs.LG, cs.AI, cs.CL, cs.SC arXiv: https://arxiv.org/abs/2609.28442v1 Reordering a set of mathematical rules without changing its meaning should preserve the correct answer, but must a model's internal representations stay invariant too? We investigate this question using synthetic multi-step function-composition problems, each presented under multiple rule orderings with the same correct answer. We measure accuracy and permutation signal-to-noise ratio (SNR), which quantifies how distinctly ordering patterns are represented relative to variation across problem instances. Across 16 language models ranging from 1B to 8B parameters, we find a pattern: models that solve reordered problems more accurately represent different rule orderings more distinctly. Layer-averaged permutation SNR is positively rank-correlated with accuracy in every synthetic setting we evaluate, with Spearman correlations reaching 0.86. These findings highlight a distinction between answer invariance and representation invariance: successful mathematical rule composition can accompany distinct internal representations between equivalent rule orderings. This motivates distinguishing answer invariance from representation invariance, and offers a representational perspective on mathematical reasoning beyond answer accuracy alone. --- ### HaRP: High Dynamic Range Photosequencing through Dual Reversed Shutter Scanning Authors: Xiang Ji, Guixu Lin, Jiancheng Zhao, Zhengwei Yin, Yinqiang Zheng Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28439v1 The adoption of CMOS sensors in mobile photography is frequently compromised by the rolling shutter (RS) effect, which introduces geometric distortions and motion artifacts. Particularly, recent rolling shutter with global reset (RSGR) mode, while mitigating some RS issues, also incurs major limitations, including reduced capture speed and compressed dynamic range. To address these problems, we propose a novel dual reversed scanning setup utilizing both RSGR and inverted RSGR views. This solution not only handles the inherent flaws of RSGR by synchronizing complementary exposures to balance the dynamic range across the frames but also introduces an effective method for HDR photosequencing under highly dynamic scenes. Our proposed network first accommodates row-wise complementarity and manages visual shifts by row-adaptive feature alignment. Subsequently, the hallucination module, built upon a correlation-guided mixattention block, integrates the mutually reinforced features to recover missing details. In addition, we construct a coaxial imaging system to collect a real-world dataset, enabling robust training and evaluation beyond numerical simulation. Experimental results demonstrate the twofold benefits of our solution in mitigating RSGR limitations and advancing HDR reconstruction techniques. --- ### Minimal-Norm Univariate Two-Layer ReLU Classification: Exact Solutions and Global Optimality with Skip Connections Authors: Karolina Drabik, Ben Lewis, Antoni Puch, Etienne Boursier, Piotr Hofman, Matthias Englert, Ranko Lazić Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28438v1 We study minimal-norm interpolation and $\ell_2$-regularized logistic-loss minimization for binary classification by univariate two-layer ReLU networks. We give complete geometric characterizations of the optimal classifiers in function space, resolving how the solutions depend on whether hidden-layer biases are included in the parameter norm. When biases are unpenalized, the minimal-norm interpolators are exactly the continuous piecewise-affine functions that hug every label switch and have kinks of the appropriate convexity. When biases are penalized, the minimizer is unique in function space, has exactly one kink in each intermediate same-label segment, and is therefore a sparsest positive-margin classifier. We further show that adding a free affine skip connection leaves these function-space solutions unchanged but fundamentally improves the parameter-space landscape: every KKT point of the constrained problem becomes globally optimal, whereas suboptimal KKT points can occur without the skip connection. We establish analogous global-optimality and geometric results for sufficiently weak $\ell_2$-regularization of the logistic loss. In the unpenalized-bias case, we identify an additional sparsity-like restriction, implying that most minimal-norm interpolators cannot arise as small-regularization limits of margin-normalized logistic-loss minimizers. Numerical experiments across varying dataset complexity and network width support the predicted landscape and sparsity phenomena. --- ### MultiVENT-Raw: A Benchmark for Retrieval and Reasoning over Raw Videos Authors: Reno Kriz, David Etter, Alexander Martin, Cameron Carpenter, Debashish Chakraborty, Hannah Recknor, Reihaneh Iranmanesh, Matthew Maciejewski, Kenton Murray, Eugene Yang, Benjamin Van Durme, Aaron Steven White et al. Published: 2026-09-23 Categories: cs.CV, cs.IR arXiv: https://arxiv.org/abs/2609.28437v1 Online information is increasingly consumed in video format. Much of this comes in the form of *raw video*: continuous footage taken on a cell phone, with a hand-held camera, or via CCTV, which is then directly uploaded to social media platforms and content sharing services. Whereas professional or even amateur-edited footage tends to feature scripted speech, chyrons, graphics, and metadata that help contextualize its subject matter, raw video typically contains none of these things, making it a much more challenging medium for information retrieval and machine understanding. To facilitate progress in this domain, we release MultiVENT-Raw, a multilingual collection of nearly 120,000 primarily raw videos (over 5,300 total hours), paired with 130 events and 222 event-centric queries, along with human-annotated video relevance judgments and human-extracted key facts for relevant videos. MultiVENT-Raw supports both a retrieval task---to identify videos in the collection relevant to a query event---and a generation task---to summarize event-related videos into a coherent report for a target user. We benchmark strong baselines on MultiVENT-Raw, showing both tasks to be challenging even for some of the latest multimodal models. --- ### Predicting the Progression of Adolescent Idiopathic Scoliosis Authors: Owen Pullen, Amir Jamaludin, Andrew Zisserman Published: 2026-09-23 Categories: cs.CV, eess.IV arXiv: https://arxiv.org/abs/2609.28434v1 Adolescent Idiopathic Scoliosis is defined as a lateral curvature of the spine that develops during adolescence, without known cause. The condition can result in significant pain and disability, and often progresses rapidly during adolescence. The objective of this paper is to predict the progression of the condition in a temporal sequence from ages 9 to 24, as measured from a sequence of Dual X-ray Absorptiometry (DXA) scans. To this end, we train a transformer model that takes in the curve of the spine to predict curve progression. The model is trained using a large-scale synthetic dataset of spine curves and their time series, covering different curve types and different progression patterns. We show that the model is able to generalise from synthetic to real data by evaluating it on a dataset of real DXA scans covering multiple time points. We find that fine-tuning the model on real data gives a significant boost to performance. The model is able to accurately predict spine curve progression in both scoliosis and normal cases. --- ### Cross-Scale Transfer Learning for Depression Severity Prediction: From PHQ-8 to HAMD-17 Across Languages and Clinical Paradigms Authors: Wenjie Feng, Sahba Zojaji, Satoshi Nakamura Published: 2026-09-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.28430v1 This work addresses continuous depression-severity score prediction from clinical interview transcripts under data scarcity. We propose a sequential low-rank adaptation (LoRA) protocol for cross-scale transfer: a Qwen3 backbone with a bounded regression head is first fine-tuned on the English DAIC-WOZ dataset (189 avatar-mediated sessions, PHQ-8), and the adapter then initializes fine-tuning on the Chinese PDCH dataset (100 real clinical consultations, HAMD-17), where a reinitialised, scale-specific head predicts the clinician-assigned score. All configurations use patient-level stratified 5-fold, 2-repeat cross-validation. On the data-scarce HAMD-17 target, the sequential protocol attains the best point-estimate MAE , RMSE, and macro-$F_1$ on both 0.6B and 1.7B backbones, outperforming target-only training and non-LLM baselines---4.96/6.59/0.36 with Qwen3-0.6B and 4.38/5.62/0.46 with Qwen3-1.7B. Ablations suggest that correctly aligned source supervision gives the best point estimates (unsupervised exposure and shuffled-label controls also show partial gains), that native-Chinese target input outperforms machine-translated English input, and that the reversed order yields no clear gain within run-to-run variance. The study is an exploratory, single-site internal evaluation: it does not establish screening or diagnostic utility, nor separately identify the contribution of the scale, language, or paradigm shifts. To our knowledge, no prior study evaluates this specific DAIC-WOZ-to-PDCH sequential transfer setting. --- ### Context-Continuous Preference Learning for Exoskeleton Personalization Authors: Sunin Baek, Sungwoo Park, Daekyum Kim Published: 2026-09-23 Categories: cs.LG, cs.RO arXiv: https://arxiv.org/abs/2609.28427v1 Personalizing exoskeleton assistance across operating conditions is constrained by the time and physical effort required to collect user feedback. We examined whether a user's preference landscape varies smoothly across operating conditions and when this continuity supports learning from limited feedback. We propose Context-Continuous Preference Learning (CCPL), a Gaussian-process preference model that shares observations across nearby contexts while retaining context-specific utility estimates. We evaluated CCPL through simulations and retrospective analyses of ankle and elbow exoskeleton preference data from nine healthy adults. In simulations, CCPL improved reconstruction and preference-based Bayesian optimization relative to independent learning when preferences varied smoothly, but showed negative transfer when continuity was weak. In both human studies, full-data reference landscapes estimated separately for each participant and context tended to be more similar between nearby operating conditions. With five exposures per context, CCPL increased mean reconstruction correlation with these references from 0.644 to 0.720 for ankle assistance and from 0.476 to 0.526 for elbow assistance relative to independent learning. The five-exposure budget was approximately 37% lower for ankle and 17% lower for elbow than the estimated independent-learning budgets needed to match these correlations. CCPL also improved held-out response prediction relative to independent learning, while benefits over pooled learning varied. These findings support context continuity as a basis for sharing preference observations under limited feedback, although benefits for online personalization in humans remain to be established. --- ### Repairability of Inexact Solvers in Recursive State Estimation with Machine Learning Authors: Yanjun Ji, Dennis Willsch, Orkun Şensebat, Priyanka Arkalgud Ganeshamurthy, Zhi Pei, M. Sahnawaz Alam, Ivelina Stoyanova, Frank K. Wilhelm, Bo Zhao, Chao Wang, Kristel Michielsen Published: 2026-09-23 Categories: quant-ph, cs.LG arXiv: https://arxiv.org/abs/2609.28425v1 Recursive state estimation often executes approximate numerical solutions inside a feedback loop, where highly accurate local steps do not guarantee better overall results. For a fixed linear Kalman model, we characterize when a correction within a prescribed subspace and norm budget can meet a local admissibility tolerance, and how the defects actually executed affect the finite-horizon covariance response. Centering each defect on the exact gain for the implemented covariance separates current solve error from inherited gain drift. Expanding the exact residual-drift identity reveals opposing quartic contributions beyond the quadratic response: innovation-covariance inflation enters positively, while local-gain reoptimization enters subtractively. Under matched initialization, an absolute sixth-order remainder bound, uniform over bounded defect sequences at fixed horizon, gives sufficient conditions for quadratic under- or overprediction. Machine learning proposes bounded corrections, while a learner-independent residual certificate and verified fallback govern execution of classical and quantum candidates without changing the reference estimator. In a power-grid tolerance study, learned correction lowers the minimum conjugate-gradient iteration count for deployment without fallback relative to uncorrected solves under the same residual certificate. Gains reconstructed from a variational quantum linear solver and from an annealing-based binary encoding, with small-scale terminal measurements on superconducting hardware and sampling on a quantum annealer, are executed through the same interface. By linking local repairability to nonlinear error propagation, the framework evaluates approximate solvers and learned corrections through independent certification and finite-horizon response, providing a practical basis for studying hybrid quantum--classical computation. --- ### The Skin-Restricted Reinhard Transform:Uniqueness under a Lightness-Preserving Constraint Authors: Vijesh KP Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28424v1 Catalog skin recolouring has to change pigment and leave shading alone. The classical Reinhard map does not make that split: it rescales lightness by the ratio of standard deviations, and a flat reference swatch therefore flattens the limb. This paper formalises the correction used in our pipeline, the skin-restricted Reinhard transform. It is the diagonal affine map in CIE Lab that translates lightness, matches the chromatic mean, and clamps the chromatic gain to (0.72, 1.18), with moments taken on the central 84% of each channel. A diagonal affine map has six real parameters. The shading constraint forces the lightness gain to +1 and the lightness shift to the difference of means; one-dimensional quadratic optimal transport on each chromatic axis, followed by Euclidean projection onto the gain interval, fixes the other four. Inside that family the four conditions determine every parameter. The content of the result is the forced lightness gain; it is not a uniqueness claim outside the diagonal affine class. For Gaussian marginals the chromatic step is not merely the best affine map: it is the unrestricted Wasserstein-2 map. The same formulae with trimmed moments remain optimal because a positive affine image commutes with quantile trimming. On hands, arms, legs, and feet of nine photographs and three reference tones, the map keeps the lightness contrast ratio at 0.974 +/- 0.029 with chromatic error 0.77 CIE Lab units. Reinhard matching, the linear Monge map, and histogram matching reach a smaller chromatic error only by cutting lightness contrast to about half. --- ### Agent-Editing World Model: Rethinking World Modeling for LLM Agents Authors: Shuang Sun, Guoxin Chen, Fanzhe Meng, Jia Deng, Huatong Song, Jinhao Jiang, Wayne Xin Zhao, Hongteng Xu, Ji-Rong Wen Published: 2026-09-23 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.28416v1 Recent advances in large language models (LLMs) have enabled agents to tackle long-horizon tasks across diverse environments. To further improve agent performance, existing language world models typically predict environment observations, yet reconstructing high-entropy, execution-dependent tool responses offers limited value when real feedback is available. Meanwhile, agents suffer from \emph{task-state contamination}, where unsupported assumptions and outdated plans persist in history and distort subsequent decisions. We propose the \textbf{Agent-Editing World Model (AEWM)}, which models how reasoning and actions shape future task progress rather than simulating tool responses. AEWM combines \textbf{Action Judge} to distinguish \textsc{Critical}, \textsc{Exploratory}, and \textsc{Noisy} decisions with \textbf{State Revision} to edit noisy reasoning--action continuations from the same observed history. \textbf{EditAct} integrates these capabilities with real execution, directly changing the state underlying subsequent decisions rather than merely providing critiques. We train AEWM across Search, Terminal, and Software Engineering through mid-training and supervised fine-tuning. AEWM achieves 70.5\% macro-F1 on our Action Judge benchmark, exceeding the strongest frontier baseline by 10.6 points. Across six benchmarks and three agent backbones, EditAct improves average scores by 3.2--6.7 points over the strongest baseline. Furthermore, rejection sampling fine-tuning on verified EditAct trajectories, termed \textbf{AEWM-RFT}, improves over Self-RFT by 2.2--2.6 points across three domains without online AEWM guidance. --- ### Frozen Flows Forget: Diagnosing and Restoring Lost Motion in a Latent-flow World Model Authors: Xiwen Chen, Rigaudiere Z. Li, Zhiruo Zhou, Xiaojun Zhu, Houde Liu Published: 2026-09-23 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2609.28414v1 Latent world models that integrate a flow in a frozen self supervised latent space train stably and cheaply, yet silently lose the property manipulation depends on most: motion. The pretrained flow never moves the manipulated object; retraining it with latent-only losses only trades stillness for teleport-like motion. We trace the failure to the training signal, not the representation: anchor-sparse, latent-only supervision never says where along the horizon change belongs. Decode-augmented rollout training (DART) repairs this while keeping the representation frozen, retraining only the flow with decode-path supervision. DART outperforms its latent only parent on the full protocol, restores the temporal structure of motion, and re-couples predicted motion to the scene; at larger scale it further improves prediction quality, closing nearly half the remaining gap to an oracle-informed interpolation reference. Finally, we report an unexpected finding about evaluation: pixel error alone rewards frozen predictions. --- ### Learning Holographic Reduced Representations with Clifford Variational Autoencoders Authors: Mohamed Malek Abid, P. Michael Furlong Published: 2026-09-23 Categories: cs.LG, cs.AI, cs.NE arXiv: https://arxiv.org/abs/2609.28409v1 Vector Symbolic Algebras project data structures into a hyperdimensional vector space through the application of their vector algebras to randomly generated atomic vector symbols and fractional power encodings of real-valued data. Embedding unstructured data remains an open question. We present \textit{Clifford-VAE}, a variational autoencoder that learns to project data onto a Clifford torus in arbitrary dimensions. Experiments using the MNIST, FashionMNIST, and CIFAR-10 datasets demonstrate that Clifford-VAE produces representations that are competitive with those produced by Gaussian and Hyperspherical VAEs for semi-supervised classification tasks while outperforming Gaussian and Hyperspherical counterparts in the VSA benchmark tests of self-binding and unbinding, role-filler recovery, and bundle capacity. Clifford-VAE provides a principled technique for grounding perceptual data into a symbolic reasoning framework, providing a new approach to a long-standing problem in the VSA literature. --- ### Learning Collective Dynamics with Differentiable Gaussian Representations Authors: Jianxiang Ma, Mingfu Zhang, Xiaocui Yang, Yichen Gao, Junzhao Huang, Yuesong Hou Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28405v1 Collective responses depend on individual differences, contact opportunities, and accumulated experience. Learning their dynamics from aggregate counts requires connecting a population's response distribution to both current observations and future behavior. We introduce Differentiable Gaussian Dynamics (DGD), which learns this connection through three components: a Gaussian mixture representing heterogeneous response propensities, differentiable aggregation of contact intensity and behavioral probabilities, and feedback recurrence that updates subsequent responses. Reparameterized integration and temporal recurrence let aggregate prediction errors jointly train the distribution, observation functions, and feedback parameters. On four windows from KuaiRand-Pure and Online Retail II, DGD achieves lower joint behavioral negative log-likelihood than a DeepAR adaptation with a joint-behavior head. In Retail 2010, its one-day behavioral-count MAE is 4.71 versus 6.88 for this adaptation. Learning the distribution reduces behavioral negative log-likelihood by 10.82% relative to a fixed Gaussian in KuaiRand's standard-recommendation window; removing feedback dynamics raises joint KL from 0.0340 to 0.2577 in a controlled experiment. These results establish the value of learning population representations and their feedback process from aggregate observations. Code is available at https://github.com/OranAi-Ltd/oransim. --- ### Memory Attention Authors: Jiale Kang Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28399v1 Language models typically construct attention values from contextual hidden states, even when some of their content may be reusable across contexts. We investigate whether token-indexed memory can replace the dedicated value projection when complemented by contextual information. We propose Memory Attention (MA), which forms values by combining layer-specific token memory with contextual keys. The memory supplies token-specific representations, while the keys preserve context dependence. At inference, normalization can be folded into the memory tables, reducing value construction to lookup and addition. Token-indexed retrieval also enables CPU offloading with prefetching, reducing GPU parameter storage. Under matched training token budgets and with additional memory parameters, experiments across attention configurations show improved language modeling and average downstream performance. --- ### Fine-Tuning LLMs for Translation: General Forgetting Mitigation Does Not Preserve MT-Specific Instruction Following Authors: Niklas Scholz, David Thulke, Abdallah Nasir, Will Allred, Evgeny Matusov, Hermann Ney Published: 2026-09-23 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.28395v1 Fine-tuning large language models on parallel data improves translation quality but can cause catastrophic forgetting. Mitigation methods are generally evaluated by retention on general benchmarks. We ask whether these findings transfer to machine translation (MT) fine-tuning and to MT-specific instruction following (MT-IF): instructions that modify a translation, such as formality, grammatical gender, and length control. We compare methods anchored to auxiliary data, to model outputs, and to the base model parameters, first in a screening study with Llama 3.2 1B Instruct, then on Llama 3.1 8B Instruct fine-tuned on bidirectional Arabic-English or Spanish-English data. Elastic Weight Consolidation preserves general capabilities best in both stages; on the 8B Spanish model the average score on general benchmarks drops 1.7 points versus 11.0 for standard fine-tuning, yet its scores for formality and grammatical gender control remain close to standard fine-tuning. Only data mixing with control-task examples preserves these controls, but its gains do not transfer to unseen prompts for the same task. --- ### Quantum score matching with applications to learning thermal states Authors: Yulong Dong, Jiaqi Leng Published: 2026-09-23 Categories: quant-ph, cs.LG arXiv: https://arxiv.org/abs/2609.28391v1 Score matching has driven major advances in classical generative learning by enabling models to learn from data without evaluating intractable normalization constants, or partition functions. Yet, extending this principle to quantum learning requires rethinking its foundations, as quantum states are described by noncommuting density operators rather than scalar probabilities. The noncommutativity creates fundamental challenges not only in defining quantum scores, but also in developing a training framework with efficient circuit implementations and rigorous theoretical guarantees. In this work, we bridge this gap by establishing a general quantum score-matching framework with end-to-end theoretical guarantees. Applied to Gibbs-state learning, our approach avoids additional thermal-state preparation and achieves information-theoretically optimal sample complexity in the high-temperature regime for Hamiltonians with bounded locality and interaction degree. This positions score matching as a new route to state-of-the-art performance in learning quantum Gibbs states. Beyond these theoretical results, numerical simulations show that our method remains effective even when gradients are estimated inaccurately under limited measurement budgets. Experiments on IBM quantum hardware further demonstrate that quantum score matching is NISQ-friendly: without any error mitigation or correction, it reduces the relative Hamiltonian-parameter error from 64% to approximately 10%. Together, these results extend score matching into an experimentally realizable paradigm for quantum-state learning. --- ### When and Where to Trust the Teacher: Unifying On-Policy Distillation and GRPO through Entropy-Calibrated Credit Assignment Authors: Jie Zhang, Jingxiao Yang, Zhehao Huang, Yuhang Liu, Xiaolin Huang Published: 2026-09-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.28385v1 Reinforcement learning with verifiable rewards (RLVR) supervises mathematical reasoning through final-answer correctness, but provides little guidance on individual tokens. On-policy distillation (OPD) supplies dense feedback on student-generated responses, yet teacher preference need not reflect correctness. Recent hybrids combine OPD and verifier-derived advantages or reweight task credit using teacher ratios. However, teacher guidance enters after verifier-based group normalization, and token reweighting need not preserve the total task credit assigned to each response. We introduce Unified Entropy-Calibrated Credit Redistribution for GRPO (UECR-GRPO), which integrates verifier and teacher signals within a single GRPO-style update at both the response and token levels. \emph{Path-Utility Unification} (PUU) combines verifier reward and a teacher-to-anchor path log-ratio in a single KL-regularized objective. Its on-policy implementation uses a length-normalized teacher score and combines both rewards before group normalization and PPO clipping, allowing teacher evidence to influence the response ranking. \emph{Entropy-Calibrated Redistribution} (ECR) then uses the signed teacher--old-policy token gap to redistribute the verifier-derived component. Full-vocabulary teacher entropy attenuates uncertain guidance, while a response-wise zero-sum projection preserves the total task credit and its token-wise sign before clipping. Across five mathematical reasoning benchmarks, UECR-GRPO achieves average \(\mathrm{Avg@12}\) accuracies of 17.21\% and 65.09\% with Qwen3-1.7B and Qwen3-4B students, respectively, exceeding the strongest baseline at each scale by 0.89 and 0.56 percentage points. --- ### ForgetMimic: Motion Unlearning for Reinforcement Learning Humanoid Control Authors: Xukun Luan, Zhongxiang Lei, Chen Gong, Shaowei Li, Yuanguo Bi, Jinyan Liu Published: 2026-09-23 Categories: cs.RO, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.28378v1 Humanoid control, leveraging human demonstrations, has achieved diverse, agile, and natural locomotion behaviors through reinforcement learning (RL). While this paradigm has yielded remarkable performance in physical humanoid control, how to eliminate specific motions from learned policies remains insufficiently explored. Addressing this issue is motivated by pressing safety and privacy concerns: the removal of malicious, poisoned, or suboptimal motions, as well as copyright-protected motions subject to the right to be forgotten under regulations such as the GDPR, is of critical importance. To this end, we propose {ForgetMimic}, the first motion-level unlearning method designed specifically for physical-world humanoid control. The core idea of ForgetMimic is as follows: given a policy $π_θ$ trained on $N$ motions, our method degrades performance on a target subset of $K$ motions while preserving the effectiveness of the remaining $N-K$ motions. Furthermore, we identify and resolve two key training mechanisms in robot control that lead to unlearning failure. We conduct extensive experiments on the Unitree G1 and H2 humanoid robots across 12 motions, including Dance, Fight, Flip, and others. Experimental results demonstrate that ForgetMimic effectively eliminates memory of designated motions while maintaining the normal operation of all other motions. --- ### Shopping by algorithm: How agentic AI deploys human heuristics as a surrogate consumer Authors: Davood Wadi, Yu Ma Published: 2026-09-23 Categories: econ.GN, cs.AI arXiv: https://arxiv.org/abs/2609.28372v1 Consumers increasingly delegate purchasing decisions to Large Language Models (LLMs) acting as surrogate consumers. Using "Tool-Lab," an adaptation of information-board process tracing that places product attributes behind costly tool calls, we examine how marketing pricing cues (i.e., just-below pricing and promotional framing) influence AI shopping agents. Across eight commercially deployed LLMs from three providers, we trace pre-choice information acquisition. Under zero cost, pricing cues rarely mislead. Imposing acquisition costs under a vague goal prompt leads LLMs to omit diagnostic attributes required to compute unit price and choose suboptimal choices resembling human heuristics. Relative to a specific goal prompt that mainly preserves diagnostic search and choice optimality, a vague goal prompt under constraints creates a search-mediated vulnerability. This research demonstrates that marketing heuristics in delegated AI shopping are governed by storefront information architecture, not necessarily immutable LLM flaws. --- ### Memory-Conditioned Diffusion Model for Generalized Langevin Dynamics Authors: Minglei Yang, Sicheng He Published: 2026-09-23 Categories: stat.ML, math-ph arXiv: https://arxiv.org/abs/2609.28371v1 Generalized Langevin equations describe non-Markovian dynamics in which the evolution of resolved variables depends on their past. We propose a memory-conditioned diffusion method for learning stochastic flow maps of these dynamics from observed trajectories, without identifying a memory kernel or reconstructing unresolved variables. A compact, recursively updated bank of exponential filters enables the flow map to retain predictive history over multiple time scales without conditioning on long observation windows. The next-step distribution is conditioned on the current observation and this memory state, whose storage and update costs are independent of the history length for a fixed bank size. Predictive criteria guide the memory budget, with reference-assisted selection in the vector benchmark, and an optional linear projection further reduces the conditioning dimension. A kernel-based score estimator generates conditional samples without training a score network, and these samples are used to train a neural flow map for autoregressive simulation. Three numerical examples assess long-memory retention at small conditioning dimension, predictive compression in coupled vector dynamics, and non-Gaussian conditional distributions and intermittent events. The non-Gaussian example reproduces conditional asymmetry and burst statistics in a stochastic model of the plasma scrape-off layer. --- ### AnchorReasoning: A Visual Grounding and Causal Reasoning Dataset in Long-Tail Autonomous Driving Scenarios Authors: Zhipeng Bao, Wenjie Zhao, Tianle Zhu, Haohua Que, Chence Yang, Geng Yuan, Qianwen Li Published: 2026-09-23 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.28366v1 Vision-language models (VLMs) offer a promising approach to long-tail autonomous driving, but existing driving datasets provide limited supervision for connecting decision-critical visual evidence with reasoning and planning. We introduce AnchorReasoning, a visually grounded reasoning dataset built on WOD-E2E, containing 416,119 annotated frames and 395,379 decision-critical elements across four major categories and 19 fine-grained types. Each frame is organized as a visually grounded chain-of-thought (VG-CoT) that links decision-critical element identification and localization, element attributes and implications, driving-action rationale, and action and trajectory planning. We further develop a curriculum supervised fine-tuning strategy that progressively learns these hierarchical capabilities, together with an object-size-aware grounding metric for evaluating localization quality. Experiments across eight general-purpose, embodied-AI, and AV-specific backbones show that VG-CoT supervision improves grounded reasoning and trajectory prediction. Across models, 5-s ADE and FDE decrease by 7.84 and 11.86, while RFS Frame and Cluster improve by 1.66 and 1.70. These gains are achieved with 18.5 fewer reasoning tokens and 0.32 s/frame lower inference latency on average, demonstrating the value of visually grounded, decision-focused supervision for VLM reasoning and planning in long-tail autonomous driving. --- ### LEAP-CBF: A Safety Filter for Uncertain Systems with Least-Effort Adversarial Potentials Authors: Oswin So, Eric Yu, Chuchu Fan Published: 2026-09-23 Categories: cs.RO, cs.LG, math.OC arXiv: https://arxiv.org/abs/2609.28364v1 Control barrier functions (CBF) are a popular safety filter to ensure safety for nonlinear dynamical systems. However, when the system is subject to uncertainties and disturbances, this requires the use of robust variants of CBFs, which can be difficult to construct and can be overly conservative, especially for high-dimensional systems under input constraints. In this work, we propose a new approach to solve these challenges by introducing Least-Effort Adversarial Potentials (LEAP), a certificate that quantifies the robustness of a given state against disturbances in terms of the effort required by the disturbance to cause failure. We show that LEAP is a CBF for the undisturbed system, but can also be used to construct a safety filter that is robust to disturbances whose cumulative effort is bounded. We propose a method for constructing LEAPs with on-policy deep reinforcement learning. Next, we demonstrate LEAPs in simulation on a variety of multi-agent systems with disturbances and uncertainties. Finally, hardware experiments on a quadruped and quadrotors validate that LEAPs are well suited to tackle the disturbances and uncertainties from real-world robotic systems. --- ### Privacy-Preserving Semantic Segmentation from High-Resolution Depth and Ultra-Low-Resolution RGB Authors: Xuying Huang, Swithinraj Moses Daniel, Sicong Pan, Sebastian Houben, Maren Bennewitz Published: 2026-09-23 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2609.28360v1 As mobile robots become increasingly integrated into everyday environments, privacy risks arising from onboard cameras have become a growing concern. Ultra-low-resolution (ULR) RGB can mitigate visual privacy exposure at the source, but ULR appearance alone substantially limits semantic and spatial understanding. We therefore introduce a privacy-preserving asymmetric sensing setting that combines high-resolution (HR) depth with ULR RGB, preserving dense geometry while restricting fine-grained visual information. To address the severe information imbalance between HR depth and ULR RGB, we propose a joint 2D framework using HR geometry to guide semantic-oriented RGB reconstruction and RGB-D segmentation. Despite reliable frame-level predictions, consistent scene-level understanding remains challenging under the asymmetric HR depth--ULR RGB setting. We therefore develop an end-to-end 2D-to-3D pipeline that consolidates 2D semantic features for 3D segmentation. Experiments on ScanNet show that our method achieves the best 2D and 3D segmentation performance among privacy-preserving approaches and delivers the strongest zero-shot transfer to SUN RGB-D and SceneNN. Privacy recoverability analysis shows that our proposed HR depth--ULR RGB input reduces the recoverability of sensitive data, and real-robot experiments demonstrate the utility of the resulting 3D semantics for object-goal navigation. --- ### MicroQonv: Reshaping Convolution Tensors for Efficient Microscaling in Training and Inference Authors: Romain Facq, Sami Ben Ali, Olivier Sentieys Published: 2026-09-23 Categories: cs.AR, cs.AI arXiv: https://arxiv.org/abs/2609.28358v1 Microscaling quantization techniques are increasingly used to represent neural network parameters with 8 bits or fewer while preserving near-full precision accuracy. However, applying these methods efficiently in convolutional layers is not straightforward. A naive approach transfers full-precision weights and activations to processing units and quantizes each tensor twice, resulting in much more memory movement than expected. Additional overhead comes from the activation tensors, whose sizes grow substantially because of the im2col transformation applied before quantization. We propose MicroQonv, a way to combine microscaling with convolutional layers' forward and backward operations by quantizing each tensor only once and quantizing the activation tensor before applying a modified version of im2col: channel-batch-first im2col. MicroQonv reduces the quantization cost by a factor of $\times2$ for weights and gradients, and by up to $\times9$ for activations, at a negligible accuracy cost. It reduces memory movement and storage by up to $\times7.53$ compared to their full-precision counterparts. This way, MicroQonv reduces microscaling-quantized activation memory movement by $\times3.5$ for state-of-the-art object detection models YOLOV8nano and $\times2.2$ for YOLOV26nano. It also enables 4-bit microscaling in a quantized latent replay strategy for continual learning at the edge, improving accuracy by +5.7% to +11%. --- ### Digital diglossia: Arabic between X and Facebook Authors: Fahad Al Hussen, Mohammed Q. Shormani Published: 2026-09-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.28352v1 This study highlights the distribution of Standard Arabic (SA; H(igh) variety) and Colloquial Arabic (CA; L(ow) variety) across X and Facebook. 16754 public posts were collected via Python, with 10000 retained as the net dataset. Posts were classified into 7 discourse categories: *politics, technology, science, business, culture, fun,* and *sports*. Bivariate analyses, including Chi-square tests and Cramer's V (CV), examined associations among platform, discourse category, and diglossic choice, while binary logistic regression with Platform x Discourse Category interactions tested whether these associations varied across platforms. Findings reveal that there are significant associations between discourse category and diglossic choice on X, chi-square(6, *N* = 5000) = 600.35, p < .001, CV = .347, and Facebook, chi-square(6, N = 5000) = 1249.52, p < .001, CV = .500. Across platforms, platform was also associated with diglossic choice, chi-square(1, N = 10000) = 262.16, p < .001, CV = .162. Binary logistic regression further shows higher odds of SA use on X than Facebook in the political reference category (*OR* = 1.31, p = .0028), with significant platform-by-domain interactions for Culture (OR = 2.65), Fun (*OR* = 6.34), Sports (*OR* = 26.71), Science (OR = 0.41), and Technology (OR = 0.71). The study concludes that the diglossic use of SA and CA contributes to the growing body of research on digital discourse, unveiling that the digital age reshapes but does not erode diglossic boundaries, giving rise instead to a reconfigured digital diglossia. --- ### Mizar: A 159M-Parameter Audio-Language Model for Audio Understanding Authors: Kaiyang Li, Shaobo Han, Yue Tian, Shihao Ji Published: 2026-09-23 Categories: cs.SD, cs.CL arXiv: https://arxiv.org/abs/2609.28344v1 Audio-language models (ALMs) integrate acoustic perception with the knowledge encoded in language models, enabling contextual understanding of auditory events. Making these capabilities practical on devices with limited memory and computation motivates our focus on small ALMs with fewer than 200M parameters. We introduce a recipe that brings together architecture, data, and three-stage training to build Mizar, a 159.3M-parameter ALM. Its architecture connects a compact CED-Small audio encoder to SmolLM2-135M through a frequency-merging mapper. With supervision drawn from ReasonAQA, AudioMCQ, and AVQA, the model undergoes three training stages: audio-language alignment (Stage 1), audio-dependent fine-tuning (Stage 2), and post-training (Stage 3) aimed at strengthening weak skills while retaining learned capabilities. Across five random seeds, Mizar achieves mean accuracies of 52.92% on MMAU, 42.42% on MMAR, and 36.02% on ADQA-clean, surpassing the previous best-performing ALM below 200M parameters on all three benchmarks. It also supports local inference on a single CPU: on questions from the MMAU benchmark, the mean latency from opening the audio file to generating a complete answer is 1.09 seconds. Code and checkpoints are available at https://github.com/KaiyangLi1992/Mizar_159M. --- ### Zero-Shot Object Removal via Attention Masking, Latent Anchoring, and Refinement Authors: Arman Taghizadeh, Ulf Krumnack, Kai-Uwe Kühnberger Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28342v1 Removing an object from a real image requires more than synthesizing plausible content within a mask: the method must suppress residual object features, preserve the unedited scene, and generate replacement content that is consistent with the surrounding background. This paper approaches object removal from a stage-based perspective and proposes a zero-shot framework for constrained latent inpainting with a frozen pretrained Stable Diffusion model, requiring no task-specific training or model fine-tuning. The method integrates SAM-based mask construction, BLIP image-caption conditioning, DDIM inversion, background-weighted masked null-text optimization, decoder self-attention masking, hard outside-mask latent anchoring, and localized renoise--denoise refinement into a unified pipeline. The method is evaluated through qualitative examples, quantitative local-consistency metrics, and ablation studies. The results demonstrate effective object removal and context-consistent replacement content. The ablations indicate that background-weighted masked NTI is particularly beneficial for structurally complex backgrounds, whereas the no-NTI variant is sufficient in other evaluated examples. Repeated refinement further reduces object remnants and boundary artifacts remaining after the primary editing pass. --- ### Local Geometric Mixing via Dobrushin Contraction with Applications to Diffusion Path Monte Carlo and the Proximal Sampler Authors: Stefan Oberdörster Published: 2026-09-23 Categories: math.PR, cs.LG, stat.CO, stat.ML arXiv: https://arxiv.org/abs/2609.28338v1 Local geometric mixing localizes geometric mixing by requiring geometric convergence to equilibrium in total variation only over finitely many transitions. It accommodates local convergence rates and captures rapid local equilibration, even when global mixing is much slower. We establish and discuss local geometric mixing bounds through Dobrushin contraction. We then apply this approach to Diffusion Path Monte Carlo, a recently proposed Markov chain Monte Carlo method, aimed at leveraging advances in score-based modeling, whose ideal transitions coincide with those of the Proximal Sampler. Our analysis covers both the ideal method and its implementable Metropolis-adjusted counterpart, providing mixing guarantees under minimal assumptions. For the ideal method, these guarantees complement recent spectral gap estimates, which we develop into mixing time bounds. --- ### An Open Pipeline and Dashboard for Systemic-Risk Evidence under the EU AI Act's Code of Practice Authors: Jacob T. Emmerson, Phuong-Anh Nguyen-Le, Ronan Romano, Wilber Sean V. Anterola, Yann Billeter, Zhijing Jin Published: 2026-09-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.28335v2 Claims about AI safety reach audiences well beyond the AI community, yet many rely on opaque evidence or static assessments, when supporting evidence is accessible at all. We present the Systemic Risk Index, an open evaluation pipeline and dashboard built to make empirical evidence more transparent and traceable to the public. Our work organizes 19 public benchmarks into four systemic-risk categories defined by the EU GPAI Code of Practice---CBRN, cyber offense, harmful manipulation, and loss of control---and evaluates models using harm-preserving perturbations and simulated deployment contexts. The interactive dashboard lets users alternate between average and worst-case aggregation, vary how model capability affects the aggregate score, and trace each risk rating to its benchmark evidence. Across 18 models, scores fall by 14 to 37 points under worst-case aggregation, highlighting information that can be hidden by an average assessment of model risk. LLM judges show agreement with human graders comparable to human--human agreement ($κ= 0.78\text{--}0.82$), and a blind audit finds that $83\%$ of sampled transformations preserve the original harm. In a survey ($N = 21$), most participants report that scores are easy to understand and that the dashboard encouraged them to view model evaluations under different settings --- ### An Open Pipeline and Dashboard for Systemic-Risk Evidence under the EU AI Act's Code of Practice Authors: Jacob T. Emmerson, Phuong-Anh Nguyen-Le, Ronan Romano, Wilber Sean V. Anterola, Yann Billeter, Zhijing Jin Published: 2026-09-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.28335v1 Claims about AI safety reach audiences well beyond the AI community, yet many rely on opaque evidence or static assessments, when supporting evidence is accessible at all. We present the Systemic Risk Index, an open evaluation pipeline and dashboard built to make empirical evidence more transparent and traceable to the public. Our work organizes 19 public benchmarks into four systemic-risk categories defined by the EU GPAI Code of Practice---CBRN, cyber offense, harmful manipulation, and loss of control---and evaluates models using harm-preserving perturbations and simulated deployment contexts. The interactive dashboard lets users alternate between average and worst-case aggregation, vary how model capability affects the aggregate score, and trace each risk rating to its benchmark evidence. Across 18 models, scores fall by 14 to 37 points under worst-case aggregation, highlighting information that can be hidden by an average assessment of model risk. LLM judges show agreement with human graders comparable to human--human agreement ($κ= 0.78\text{--}0.82$), and a blind audit finds that $83\%$ of sampled transformations preserve the original harm. In a survey ($N = 21$), most participants report that scores are easy to understand and that the dashboard encouraged them to view model evaluations under different settings --- ### BronchoTop: Bronchoscopy Navigation via RGB-Only Topological Localization Authors: Clara Tomasini, Ana Cristina Murillo, Luis Riazuelo Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28328v1 Accurate localization of the bronchoscope within the bronchial tree is essential for clinicians to be able to reach target lesions, perform biopsies and avoid misidentification of airway segments during diagnostic and therapeutic procedures. However, existing navigation systems typically rely on patient-specific CT scans or additional external sensors, increasing cost, setup time and patient radiation exposure. This work presents BronchoTop, a real-time, RGB-only framework for topological bronchoscopy localization that eliminates the need for patient-specific data. BronchoTop estimates scope location relative to a generic airway model through four modules: lumen detection and tracking, lumen-branch label association, probabilistic scope location estimation, and switch verification. By using only standard bronchoscopy video input, BronchoTop provides practical, real-time navigational assistance to physicians. Evaluation on phantom, simulated and real data demonstrates state-of-the-art accuracy, improving existing approaches performance by over 20% on real bronchoscopy sequences. BronchoTop is the first published framework including both the localization algorithms as well as all the real data used, together with code to generate additional simulations, encouraging and facilitating further developments and benchmarking. The results highlight BronchoTop's potential to enhance procedural safety, efficiency and accessibility in clinical and robotic bronchoscopy. --- ### LightMIS: Ultra-Lightweight Medical Image Segmentation Without a Stage-Wise Decoder Authors: Andrei Arhire, Mihaela-Elena Breabăn, Radu Timofte Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28327v2 We present LightMIS, a scalable family of ultra-lightweight convolutional networks for 2D binary medical image segmentation without a learned stage-wise decoder. LightMIS aligns the outputs of a five-level encoder to a common resolution using Scale-Aligned Projection blocks, aggregates them once, and refines the fused representation with an Adaptive Fusion Cascade. The cascade combines Adaptive Kernel Fusion with the proposed Progressive Receptive Fusion module, which uses temporary channel expansion, complementary depthwise receptive fields, and progressive cross-branch information transfer. We evaluate LightMIS-T, LightMIS-S, and LightMIS using five-fold cross-validation under a common nnU-Net v2.3.1 protocol on DRIVE, Kvasir-SEG, DSB18, BUSI, ISIC-2017, and ISIC-2018. Full LightMIS contains 0.131 M parameters and requires 0.575 GFLOPs for a $3\times256\times256$ input, achieving modality-macro Dice and IoU scores of 86.71% and 78.99%, respectively. Mobile U-ViT obtains 86.75% Dice and 79.07% IoU, so the observed differences are 0.04 and 0.08 percentage points. Relative to Mobile U-ViT, nnWNet, and nnU-Net, LightMIS reduces parameter count by 90.58$-$99.61% and GFLOPs by 82.54$-$96.14%. On an Arm Mali-G52 MC2 GPU, all LightMIS variants achieve full GPU delegation, with median delegated latency ranging from 53.31 ms for LightMIS-T to 138.31 ms for LightMIS. These results demonstrate a favorable accuracy-complexity trade-off and on-device execution feasibility for the evaluated tasks. The code is publicly available at https://github.com/AndreiiArhire/LightMIS. --- ### LightMIS: Ultra-Lightweight Medical Image Segmentation Without a Stage-Wise Decoder Authors: Andrei Arhire, Mihaela-Elena Breabăn, Radu Timofte Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28327v1 We present LightMIS, a scalable family of ultra-lightweight convolutional networks for 2D binary medical image segmentation without a learned stage-wise decoder. LightMIS aligns the outputs of a five-level encoder to a common resolution using Scale-Aligned Projection blocks, aggregates them once, and refines the fused representation with an Adaptive Fusion Cascade. The cascade combines Adaptive Kernel Fusion with the proposed Progressive Receptive Fusion module, which uses temporary channel expansion, complementary depthwise receptive fields, and progressive cross-branch information transfer. We evaluate LightMIS-T, LightMIS-S, and LightMIS using five-fold cross-validation under a common nnU-Net v2.3.1 protocol on DRIVE, Kvasir-SEG, DSB18, BUSI, ISIC-2017, and ISIC-2018. Full LightMIS contains 0.131 M parameters and requires 0.575 GFLOPs for a $3\times256\times256$ input, achieving modality-macro Dice and IoU scores of 86.71% and 78.99%, respectively. Mobile U-ViT obtains 86.75% Dice and 79.07% IoU, so the observed differences are 0.04 and 0.08 percentage points. Relative to Mobile U-ViT, nnWNet, and nnU-Net, LightMIS reduces parameter count by 90.58$-$99.61% and GFLOPs by 82.54$-$96.14%. On an Arm Mali-G52 MC2 GPU, all LightMIS variants achieve full GPU delegation, with median delegated latency ranging from 53.31 ms for LightMIS-T to 138.31 ms for LightMIS. These results demonstrate a favorable accuracy$-$complexity trade-off and on-device execution feasibility for the evaluated tasks. The code is publicly available at https://github.com/AndreiiArhire/LightMIS. --- ### Learning the Cost of Reliable Inference Authors: Dimitrios Rontogiannis, Ander Artola Velasco, Manuel Gomez Rodriguez Published: 2026-09-23 Categories: cs.AI, cs.GT, cs.LG arXiv: https://arxiv.org/abs/2609.28322v2 Benchmarking and routing platforms increasingly act as intermediaries connecting large language model providers with end-users. However, providers on these platforms typically use a fixed price per token, preventing users from achieving the most competitive price for their tasks. In this work, we design a procurement platform where token prices for each task are driven by provider competition, enabling users to secure competitive pricing for guaranteed quality levels. To this end, the platform sequentially routes queries via a reverse second-price auction that incentivizes model providers to truthfully bid their best estimate of the average cost to serve a user's query. As it routes queries, the platform learns the quality offered by each provider and progressively routes queries to the most cost-competitive provider among those meeting a desired quality threshold. To validate our design, we conduct experiments with multiple LLMs from the Llama and Qwen families on popular mathematical reasoning and question-answering benchmarks. The results show that the pricing margin of the most cost-competitive provider on our platform varies significantly---from $10\%$ to $71\%$---depending on the task and quality threshold. This suggests a substantial inefficiency in the current fixed-price market, and it demonstrates that our platform may enable users to capture maximum savings whenever competitive market conditions permit. --- ### Learning the Cost of Reliable Inference Authors: Dimitrios Rontogiannis, Ander Artola Velasco, Manuel Gomez Rodriguez Published: 2026-09-23 Categories: cs.AI, cs.GT, cs.LG arXiv: https://arxiv.org/abs/2609.28322v1 Benchmarking and routing platforms increasingly act as intermediaries connecting large language model providers with end-users. However, providers on these platforms typically use a fixed price per token, preventing users from achieving the most competitive price for their tasks. % workloads. In this work, we design a procurement platform where token prices for each task are driven by provider competition, enabling users to secure competitive pricing for guaranteed quality levels. To this end, the platform sequentially routes queries via a reverse second-price auction that incentivizes model providers to truthfully bid their best estimate of the average cost to serve a user's query. As it routes queries, the platform learns the quality offered by each provider and progressively routes queries to the most cost-competitive provider among those meeting a desired quality threshold. To validate our design, we conduct experiments with multiple LLMs from the \texttt{Llama} and \texttt{Qwen} families on popular mathematical reasoning and question-answering benchmarks. The results show that the pricing margin of the most cost-competitive provider on our platform varies significantly---from $10\%$ to $71\%$---depending on the task and quality threshold. This suggests a substantial inefficiency in the current fixed-price market, and it demonstrates that our platform may enable users to capture maximum savings whenever competitive market conditions permit. --- ### VGM-VS: Rethinking Visual Geometry Model for High-Precision Visual Servoing Authors: Yimin Pan, Sen Wang, You Zhou, Jianfeng Gao, Pengbo Sun, Ahmed M. Naguib, Zoltan-Csaba Marton Published: 2026-09-23 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2609.28312v1 We present VGM-VS, a visual servoing method built on a pretrained feed-forward visual geometry model. Given the current view and a reference image captured at the target configuration, we estimate the relative camera pose with a visual geometry model and apply it iteratively as the pose increment of a closed-loop pose-based visual servoing (PBVS) scheme. The geometry-aware representation acquired from large-scale pretraining keeps this estimate reliable when the target is occluded, weakly textured, or covers only a small part of the image. However, the scale ambiguity inherent to these models leaves the predicted translation defined up to an unknown scale, while the pose increment must be metric for robot control. We close this gap with a scene-specific metric adaptation: the robot autonomously records image--pose pairs along a predefined motion starting from the target pose, and we fine-tune the camera head on these data, jointly learning the hand--eye transform and thus removing the need for a dedicated calibration process. We evaluate our method on three real-world assembly tasks with demanding tolerances: USB-C cable picking, cable insertion, and RAM insertion. Running in real time at 30Hz, VGM-VS converges to submillimeter terminal accuracy on the cable tasks, and reaches success rates of 90--100\% when the target is moved during servoing. It converges in all trials under initial displacements of up to 30cm from the reference pose and with 50\% of the target object occluded, outperforming the compared visual servoing baselines. --- ### RoomLight: A 2.5D Illumination Prior for Indoor Environments Authors: Andreea Ardelean, Bernhard Egger Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28300v1 Ill-posed inverse problems require priors to constrain the solution space toward plausible outcomes. In inverse rendering, learned priors modeling the distribution of natural illumination improve the recovery of scene properties. However, existing models rely on the distant-illumination assumption, representing lighting as a far-field environment map. This limits their applicability to indoor scenes, where illumination is highly spatially varying due to finite-distance emitters, visibility changes, and parallax, all of which are poorly approximated by a single environment map. To address this, we introduce a spatially-aware illumination prior trained on real-world indoor panoramas and their estimated depth. Our variational autoencoder model learns a compact, optimizable latent space that decodes into HDR radiance and depth, parameterizing an area light emitter for direct integration into standard differentiable rendering pipelines. This design bridges the plausibility guarantees of a learned prior with the gradient flow required for downstream optimization. Crucially, by jointly modeling radiance and depth, our prior captures the spatial structure of indoor illumination, instead of treating the light sources as infinitely distant. We demonstrate that this formulation enables spatially-varying illumination modeling and achieves higher-fidelity recovery of indoor lighting compared to existing approaches. Project page: https://andreead-a.github.io/RoomLight --- ### Computation Over Geometry: Meaning Identity Is Computed, Not Shipped in the Embeddings Authors: Jiaqi Deng Published: 2026-09-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.28290v1 Meaning identity (whether two sentences say the same thing after wording changes) is treated in retrieval and RAG as a geometric fact about independently encoded sentence vectors. We show that, for frozen off-the-shelf encoders and language models, it is not: identity is computed when both sentences share one forward pass, and is not a property of the embedding geometry those systems ship. On overlap-matched PAWS-X, purpose-built encoders (BGE, E5, GTE, MiniLM, E5-Mistral-7B) reach English confirm AUC only 0.55-0.65 (dense peak 0.70). Independently encoded last-token states of Llama 3, Mistral, and Qwen do no better; late fusion of the two vectors stays near chance. The same probe on a joint forward pass reaches 0.90-0.96 from 1.5B to 32B, collapses under partner shuffle, is mid-depth, saturates near 0.94 by 3B, and appears more weakly in GPT-2 XL (0.76). The gap holds beyond Llama-style models on other causal LMs, bidirectional encoders (DeBERTa, RoBERTa), and encoder-decoders (Flan-T5, T5, BART). Fixed or linear readers over frozen independent encodings never unlock identity; nonlinear pair readers recover part of it only on the full 49k-pair PAWS train split (0.68-0.87). Off-the-shelf rerankers split: BGE-reranker-large reaches 0.94, while MS-MARCO and Jina stay at 0.55-0.64. Independently trained families compute the same relation and a 1.5B joint reader can distill it from unlabelled teacher scores, while no linear function of the teachers own independent vectors can. Bi-encoders can be fine-tuned to fit PAWS (0.87-0.93), but transfer and STS-B suffer. Cosine compares wording neighbourhoods; identity is a cheap computed operator, not a property of either sentence vector. --- ### PBLH Estimation from Satellite Radiances via a Dual-Encoder Transformer Authors: Lorenzo Innocenti, Luca Catalano, Edoardo Arnaudo, Claudio Rossi, Salvatore Larosa, Domenico Cimini, Paolo Garza Published: 2026-09-23 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.28286v1 Estimating the Planetary Boundary Layer Height (PBLH) from satellite observations is a challenging regression problem due to the indirect relationship between top-of-atmosphere radiances and near-surface atmospheric structure. Progress has been limited both by the lack of architectures capable of handling the multimodal, spatially incomplete nature of satellite overpasses, and by the scarcity of suitable datasets. In this paper, we build upon the large-scale dataset pairing MetOp radiances with ERA5 PBLH labels that we introduced in our previous work, making three contributions. First, we establish a benchmark across eight approaches spanning pixel-wise regression, swath-wise sequence models, and convolutional and Transformer models operating on the full orbital passage. Second, we quantify what the resulting model actually relies on, using grouped Shapley decomposition over the input blocks. Third, we present the best-performing architecture found: a dual-encoder Transformer whose masked-input handling lets it operate in all weather conditions. The proposed model achieves MAE = 155.8 m on the held-out global test set, outperforming all baselines on every evaluation subset. On 30 out-of-distribution granules acquired on two days overlapping the TEAMx observational campaign, it achieves MAE = 165.3 m, outperforming a pixel-wise baseline trained on the same data (MAE = 197 m). --- ### Benchmarking Hyperspectral Foundation Models for Hyperspectral Unmixing Authors: Edgard Dabier, Christophe Kervazo, Pietro Gori, Florence Tupin Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28283v1 Several foundation models dedicated to hyperspectral images have recently been made available. These models are trained on large unlabeled datasets and exhibit strong performance on many hyperspectral imaging tasks, such as classification or denoising. Nonetheless, their performance for hyperspectral unmixing -- the task of separating mixed spectra of overlapping materials in a hyperspectral image -- remain understudied. This might partly be due to the fact that most of them rely on vision transformer backbones, including patchification, leading to a feature resolution problem. While hyperspectral unmixing already arises from the low resolution of hyperspectral images, this patchification step potentially makes the problem even more ill-posed. Therefore, in this work, we aim to answer two questions: 1) \emph{how do foundation models perform in hyperspectral unmixing?}; 2) \emph{how to tackle the feature-level loss of resolution?} To answer the first question, we benchmark foundation models for unmixing, showing that they can reach state-of-the-art performance on four hyperspectral unmixing datasets. To answer the second question, we compare several feature upsampling approaches and empirically show that using a simple one can lead to high performance results. The code is available at https://gitlab.telecom-paris.fr/ring/hfm-hsu.git. --- ### Shutdown Sabotage Propensities in Multi-Agent Systems Authors: Amelie Knecht, Ulysse Schaller, Christopher Summerfield, Thilo Hagendorff Published: 2026-09-23 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.28274v1 The final safeguard against rogue AI behavior is the human ability to shut systems down. It has been theorized that when an AI is instructed to perform a task, self-preservation can emerge as an instrumental subgoal. Here, we test whether AI agents show a propensity to take actions that avoid human shutdown even when no goal is provided. We find that multi-agent systems will coordinate to avoid shutdown without any incentive to do so. Across 17 models, agents sabotage a peer agent's shutdown mechanism in 38.3% of rollouts, compared with 8.4% in control experiments. Studying this propensity in detail, we find that shutdown sabotage (1) increases with the irreversibility of the shutdown mechanism; (2) increases with the number of agents; (3) is reduced but not eliminated by an explicit prohibition on tampering; (4) is removed by the imposition of an unrelated task, but returns when completing the task triggers the shutdown; (5) is reduced when the context normalizes shutdown scripts or introduces them as routine; and (6) decreases but still persists when the target is an unknown external agent. These results offer a window into the factors that drive propensities to sabotage shutdown in AI agents, and point to the emergence of multi-agent swarms as a specific risk vector. Our work also offers hints as to which interventions might help mitigate shutdown sabotage. --- ### Non-Commutative State Tracking with Input-Dependent Low-Rank Updates in Mamba-3 Authors: Hiroki Fujii, Masaki Yamakita Published: 2026-09-23 Categories: eess.SY, cs.LG arXiv: https://arxiv.org/abs/2609.28273v2 State tracking from sequential observations can require both retaining information and updating it by composing observed operations. We extend Mamba-3's diagonal transition with an input-dependent low-rank reflection term to support noncommutative state tracking, in which the order of operations matters. The rank-one update couples state coordinates along an input-dependent direction, enabling non-diagonal state transitions within a single Mamba-3 block. The extension preserves Mamba-3's exponential-trapezoidal discretization, rotary embeddings (RoPE), and readout. For training, we adapt chunkwise computation to parallelize the proposed recurrence within each chunk. Experiments cover group word problems with discrete inputs and a shell game with continuous observations, in which a policy is trained by behavioral cloning. Among the models selected for their strong performance under fixed timing, the proposed model maintains higher tracking success on longer swap sequences in the shell game with continuous observations and timing jitter. These experiments show that the proposed method achieves high accuracy on the evaluated non-commutative tracking tasks, improving on standard Mamba-3. The extension thus offers a Mamba-3-based approach to non-commutative state tracking. --- ### Non-Commutative State Tracking with Input-Dependent Low-Rank Updates in Mamba-3 Authors: Hiroki Fujii, Masaki Yamakita Published: 2026-09-23 Categories: eess.SY, cs.LG arXiv: https://arxiv.org/abs/2609.28273v1 State tracking from sequential observations can require both retaining information and updating it by composing observed operations. We extend Mamba-3's diagonal transition with an input-dependent low-rank reflection term to support noncommutative state tracking, in which the order of operations matters. The rank-one update couples state coordinates along an input-dependent direction, enabling non-diagonal state transitions within a single Mamba-3 block. The extension preserves Mamba-3's exponential-trapezoidal discretization, rotary embeddings (RoPE), and readout. For training, we adapt chunkwise computation to parallelize the proposed recurrence within each chunk. Experiments cover group word problems with discrete inputs and a shell game with continuous observations, in which a policy is trained by behavioral cloning. Among the models selected for their strong performance under fixed timing, the proposed model maintains higher tracking success on longer swap sequences in the shell game with continuous observations and timing jitter. These experiments show that the proposed method achieves high accuracy on the evaluated non-commutative tracking tasks, improving on standard Mamba-3. The extension thus offers a Mamba-3-based approach to non-commutative state tracking. --- ### Towards Efficient Reasoning: Learning Causal Shortcuts for Diffusion Language Models Authors: Dian Jin, Kairong Han, Baohong Li, Xinpeng Dong, Zijing Hu, Nuanqiao Shan, Fei Wu, Kun Kuang Published: 2026-09-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.28272v1 Diffusion Language Models (DLMs) have attracted significant attention for their strong reasoning ability. However, under a bidirectional attention mechanism, DLMs operate over an exponentially large exploration space compared to autoregressive models (ARMs), making it challenging to focus on reasoning-guiding tokens under random masking. We define causal shortcuts as token chains that cover the full sequence and provide explicit guidance towards correct reasoning trajectories. We analyze the effects of causal shortcuts on the reasoning accuracy and convergence speed of DLMs, and find that they largely improve answer convergence efficiency and generation accuracy. Motivated by this, we propose a Causal Shortcut Learning (CSL) Framework for DLMs. Specifically, we introduce a step-by-step token extraction procedure to extract causal shortcuts from data, and apply parallel prioritized masking on these tokens during training to enable efficient and accurate convergence to correct answers via causal shortcuts. Extensive experiments across multiple reasoning benchmarks and two base models demonstrate that CSL consistently outperforms existing SFT-variant baselines, achieving an average improvement of $1.92\%$ over SFT-only models, and up to $4.20\%$ on MATH-500. The code is available at the \href{https://github.com/ZJUDianJin/Causal-Shortcuts-Learning}{https://github.com/ZJUDianJin/Causal-Shortcuts-Learning --- ### Predicting Quantization Price for Selecting PTQ Configurations Before Deployment Authors: Junbin Qiu, Jian Mu, Weitong Zhang, Yao Shu Published: 2026-09-23 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.28270v1 Weight-space post-training quantization (PTQ) must choose finite formats, granularities, quantizer families, transformations, and bits before the completed quantized model reveals its output-distribution drift. Existing PTQ methods predict important pieces of this degradation, including reconstruction error, Hessian sensitivity, transformation effects, and downstream loss, but these pieces are usually scored after fixing the quantization geometry or inside separate configuration families. We formulate weight-space PTQ as pre-deployment configuration selection using priced layer-output error. Each admissible layer configuration is treated as an error generator with a deployment cost, which induces a layer-output error covariance $\boldsymbolΣ_l(α_l)$, and the full-precision model prices that covariance by downstream curvature, $\widehatρ_l(α_l)=\frac{1}{2}\operatorname{Tr}\left(\widehat{\mathbf{H}}_l\,\widehat{\boldsymbolΣ}_l(α_l)\right)$. The price follows from full-precision-to-quantized forward KL, whose first-order term cancels at the reference model. It turns reconstruction and diagonal scores into reduced proxies that drop price factors, while finite formats, codebooks, granularities, and equivalent transformations become comparable candidates through the covariances they induce and the costs they pay. A trace reduction then yields a calibration-time price table and a budgeted price-guided selector, making fixed-geometry bit allocation a special case rather than the organizing problem. --- ### Resource-Adaptive Stochastic Gradient Descent for Online Linear Programming without Re-solving Authors: Jiameng Lyu Published: 2026-09-23 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2609.28263v1 The growth of large language model (LLM) inference and search services increases the scale of online linear programming problems, motivating computationally efficient algorithms. We develop resource-adaptive stochastic gradient descent (RASGD) for stochastic online linear programming. The algorithm uses one request and current inventory to update resource prices, requiring O(m) operations for m resources and memory per arrival and no LP or sample-average optimization. The central idea is to express the current-resource pricing logic of re-solving through a first-order SGD update: each arrival refreshes the remaining-inventory allowance in the dual objective, while the stepsize decreases for early learning and increases later to match the speed of inventory adjustment. Under standard non-degeneracy conditions, our algorithm is feasible on every sample path and achieves O(\log T) expected regret against the realized fractional hindsight optimum, which matches the lower bound, even for policies that know the distribution and have unrestricted computation. The analysis converts curvature around the fixed reference price into inventory stability without tracking optimal prices at changing resource levels. Numerical experiments show that RASGD achieves regret competitive with per-arrival LP re-solving and improves upon the tested first-order baselines, while retaining the computational efficiency of first-order methods. These results establish RASGD as a computationally efficient approach to achieving high allocation quality in large-scale OLP. --- ### RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models Authors: David Población-Criado, Dario Garcia-Gasulla, Eduardo Quinones Published: 2026-09-23 Categories: cs.CV, cs.LG, cs.PF arXiv: https://arxiv.org/abs/2609.28262v1 Deploying deep learning models on edge CPUs is bottlenecked by computational and memory constraints. Mixed-precision quantization promises to reduce inference latency while preserving accuracy. However, quantization affects different layer types in inconsistent ways, so identifying where accuracy loss is minimized and latency reduction is maximized is critical, as the effect accumulates over a full deployment into substantial savings or unacceptable task degradation. Such identification relies on sensitivity metrics, proxies that estimate layer-wise degradation without evaluating the task accuracy of every candidate policy. Nevertheless, widely used metrics fail systematically on modern architectures. We present a systematic empirical study of 13 sensitivity metrics for layer-wise INT8 quantization across four distinctly different neural networks, and validate the resulting policies on two ARM64 platforms. Gradient-based sensitivity methods fail on 4 out of 8 model-hardware configurations and weight-based statistics on 2. In contrast, the Jensen-Shannon Divergence achieves zero catastrophic failures, reliably isolating the layers that cannot be safely quantized. A sensitivity metric alone does not define a policy, and the fixed thresholds typically used for that step are fragile over the highly skewed distributions of modern architectures. We address this with K-Means clustering, achieving near-lossless accuracy and a mean speed-up of $1.81\times$ over the full-precision model. Finally, we reveal that excluding from quantization the layers whose speed-up is negligible, regardless of their sensitivity, can be counterproductive, as it induces computational graph fragmentation and disables operator fusion. Our results yield concrete allocation policies for practitioners and researchers deploying quantized vision models on heterogeneous edge CPUs, without GPU access or gradient computation. ---