# Research Papers — 2026-09 · bundle page 137 of 266 > Full-text markdown of papers #6801–6850 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/137 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/136.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/138.md) ### FedeRage: Provably Convergent Agnostic Federated Learning under General Client Drift Authors: Herlock Rahimi, Dionysis Kalogerias Published: 2026-09-17 Categories: cs.LG, cs.DC, eess.SP, eess.SY arXiv: https://arxiv.org/abs/2609.21057v1 Federated learning (FL) enables collaborative model training without sharing raw data, but its performance degrades under non-IID data and stochastic client participation. Remedies built on classical Federated Averaging (FedAvg) typically presuppose that client participation probabilities are known to the server, which is rarely the case in deployed systems. We first discuss and then characterize the optimization problem that \emph{distributionally agnostic} FedAvg actually solves when participation is entirely unknown, possibly highly skewed, and of variable size across rounds: uniform aggregation is shown to minimize a well-defined stochastic objective, weighted by the participation-induced marginal, at a standard $\mathcal{O}(1/\sqrt{T})$ rate for convex and possibly nonsmooth losses. Building on this characterization, we propose \emph{Federated Risk-Averse Averaging} (\textsc{FedeRage}), a risk-averse extension of FedAvg that embeds the \emph{Conditional Value-at-Risk} (CVaR) into the local objective within a natural distributionally robust optimization (DRO) framework. \textsc{FedeRage} implicitly upweights high-loss and infrequently participating clients while adding only a \emph{single scalar per-client}, and admits an $\mathcal{O}(κ/\sqrt{T})$ rate in which the factor $κ$ is the upper bound on the ``price" of risk aversion. In contrast with aggregation-alignment schemes based on optimal transport, which require the availability distribution as an input, \textsc{FedeRage} remains agnostic to it. Several experiments on three heterogeneous benchmarks indicate consistent improvements over state-of-the-art methods in accuracy, fairness, and convergence speed. --- ### Physically Based Rendering in the Latent Space Authors: Vuk Radovanovic, Vishesh Gupta, Adrien Gruson, Binh-Son Hua Published: 2026-09-17 Categories: cs.GR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.21054v1 Image diffusion models have shown impressive image generation capabilities but are often hard to control, in contrast to classical computer graphics pipelines such as physically based rendering. However, we observe that there is a bridge between light transport phenomena and the distribution of latent space values produced by such models. Thus, we introduce physically based rendering in the feature space learned by the variational autoencoders in generative models, enabling light transport simulation in the latent space. This allows us to leverage physically based rendering techniques to output latent maps for physically guided content generation. We propose modifications to the rendering equation, which, when paired with a differentiable renderer, can yield an optimal set of scene parameters that require only minimal refinement to accurately render into the pretrained latent space. We train our method on a single rendered image, and then demonstrate the generalization of the method to scene geometry changes, lighting changes, and camera view changes. --- ### A Lightweight Plug-in Gate for Transformer-Based Time-Series Forecasters Authors: Hongkai Zhuang, Tao Huang, Chen Hou Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.21044v1 Covariate-rich time-series forecasting requires deciding how external variables enter the target forecasting path. Existing Transformer-based forecasters usually build a covariate representation and pass it to the encoder without an explicit admission stage. This paper studies pre-encoder covariate admission as an input-side interface that regulates that representation immediately before encoder processing. We implement the interface with a lightweight representation-level pre-encoder gate that assigns sigmoid scores to representation units, and we also study a usage-regularized variant that penalizes average admission. The interface is evaluated as a plug-in module for TimeXer, Inverted Transformer (iTransformer), and Patch Time Series Transformer (PatchTST) under a zero-extra-tuning protocol, where each gated model inherits the corresponding baseline configuration. Experiments on the Electricity Transformer Temperature minute-level (ETTm1 and ETTm2) datasets, Traffic, Energy, and influenza-like illness (ILI) include paired forecasting comparisons, gate-placement ablation, initialization ablation, controlled covariate-admission analysis, and a variance inflation factor (VIF)-informed permutation feature importance (PFI) diagnostic case study. In the tested settings, the gate is competitive with the corresponding baselines, and the usage penalty reduces average admission scores while keeping forecasting errors close to the unpenalized TimeXer setting. --- ### Stiefel-AdamW: Geometry-Aware AdamW for Linear Factorization Blocks Authors: Emanuele Zangrando, Marco Sutti, Francesco Tudisco Published: 2026-09-17 Categories: cs.LG, math.NA arXiv: https://arxiv.org/abs/2609.21039v1 A pervasive structural pattern in modern deep learning is the linear factorization block: a submodule of the form $W = BA$ in which two parameter matrices are multiplied directly, with no intervening nonlinearity. Such blocks appear in LoRA adapters, low-rank compressed layers, query-key products of self-attention, and share a common pathology: the factorization is non-unique, which can destabilize training and limit usable learning rates. Despite this, factorization blocks are typically optimized with standard Euclidean methods that ignore the underlying geometry. We introduce Stiefel-AdamW, a near drop-in replacement for AdamW for use wherever such blocks appear. By constraining one factor on the Stiefel manifold while leaving the other Euclidean, Stiefel-AdamW relaxes the full $\mathrm{GL}(\mathbb{R}^r)$ gauge symmetry to a compact orthogonal symmetry, ruling out factor blow-up while retaining the coordinate-wise diagonal preconditioning that gives AdamW its practical strength. Moment estimation is performed in the ambient Euclidean space, with geometry entering only through a tangent-space projection and a manifold retraction. The implementation overhead over AdamW is minimal, and we show that the resulting optimizer inherits both the stability benefits of Riemannian methods and standard convergence guarantees. We validate Stiefel-AdamW on LoRA-style fine-tuning of GPT2, ViT, and Mistral 7B and on full pretraining of GPT2 on OpenWebText, showing consistent improvements over strong baselines at essentially no additional cost over AdamW. --- ### Scaling Discovery through Test-Time Communication Authors: Jongho Park, Vasilis Kontonis, Shivam Garg, Akshay Krishnamurthy, Dimitris Papailiopoulos Published: 2026-09-17 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.21032v1 Science advances not in isolation but through collaboration, yet existing agentic systems capture little of this. Whether communicating agents help remains an open question with mixed prior results. We show that test-time communication can substantially outperform independent parallel attempts on challenging tasks, where sharing a breakthrough can push the whole group forward. We first study the effect of scaling multi-agent test-time communication, where agents have no predefined roles and communicate via a shared directory, on ARC-AGI-3, a benchmark requiring novel problem solving. We find that a team of $k$ communicating agents, team@$k$, matches the success rate of $4k$ independent agents, and this advantage grows with $k$, suggesting gains compound with scale. The effect is not merely efficiency: a task that no single agent can solve, a team of agents can solve reliably. Furthermore, these gains transfer to research-oriented tasks, given sufficient compute. On polyomino packing, communicating agents outperform best@$k$ and exceed the prior best-known score. On MNIST classifier compression, communication surpasses the best-known human solution. A team of four agents produced a 1,957-byte classifier submission achieving 99.4% test accuracy, smaller than both the best-known human solution and the best single-agent result. These gains are not unconditional. Independent agents may outperform communication when compute is limited or when a clear measure of progress is absent. However, under sufficient compute and clear feedback, multi-agent communication consistently yields stronger results. --- ### MAGIC: Marginal-Guided Compression with Optimal Transport for Efficient Visual Document Retrieval Authors: Xu Yuan, Hua Liu, Wenqi Fan, Qing Li Published: 2026-09-17 Categories: cs.CV, cs.IR arXiv: https://arxiv.org/abs/2609.21018v1 Recent visual document retrieval (VDR) systems such as ColPali use multi-vector page embeddings, in which patch-level vectors enable fine-grained evidence matching but incur substantial index storage and MaxSim scoring overhead. Post-hoc merging offers a practical route to efficient VDR by reducing this cost without retraining the retriever, but its uniform reconstruction objectives are poorly aligned with the sparse, non-uniform patch usage induced by late-interaction retrieval. Under aggressive compression, this misalignment can preserve rarely used patches while concentrating retrieval activity on too few retained representatives. To address this misalignment, we propose Marginal-Guided Compression with Optimal Transport (MAGIC), a training-free post-hoc compressor for efficient retrieval with frozen multi-vector embeddings. MAGIC derives a MaxSim-induced compression surrogate and optimizes it through a two-marginal entropic optimal-transport formulation, where a retrieval-demand source marginal prioritizes high-use patches and a balanced target marginal regularizes retained-facet usage. Across ViDoRe benchmarks, keep ratios, and retrieval backbones, MAGIC consistently outperforms strong post-hoc compressors, with particularly large gains in the aggressive-compression regime; component ablations verify the complementary effects of its two marginals. We release the code at: https://github.com/xandery-geek/MAGIC. --- ### A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks Authors: Mike Nguyen, Nicole Mücke Published: 2026-09-17 Categories: stat.ML, cs.LG, math.ST arXiv: https://arxiv.org/abs/2609.21017v1 We study data-driven early stopping for spectral regularisation methods in the classical non-parametric regression setting. Building on the discrepancy principle, we propose a multi-scale stopping rule that applies to general kernel estimators and show that, unlike previous approaches, it achieves full adaptivity over all smoothness levels in the well-specified case. A key contribution of our work is an extension based on random feature approximations, which reduces computational cost on large datasets while preserving minimax-optimal statistical guarantees. Our procedure not only selects an optimal stopping time but also provides a fully data-driven choice of the number of random features needed to achieve optimal rates. Through the established connection between random features and neural networks in the neural tangent kernel regime, our method further yields a principled, data-driven recommendation for the network width. We prove that the resulting simultaneously chosen width and stopping time allow neural networks to attain minimax-optimal learning rates without prior knowledge of smoothness or capacity parameters. --- ### Fragment-Aware Vision Transformers for Fresco-Fragment Style Classification Authors: Sara Miketek, Biagio Barchielli, Nadeem Iqbal Kajla, Sinem Aslan Published: 2026-09-17 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.21012v1 Artistic style classification is usually studied on complete artworks, where models can exploit global composition, spatial organisation, and iconographic structure. In archaeological settings, however, artworks often survive only as fragmented remains, forcing recognition from incomplete, irregular, and context-limited visual evidence. We study fresco-fragment style classification using a progressive transformer-based framework. Starting from a ViT-B/16 baseline, we introduce foreground-guided masking to suppress background-only tokens, inpainting-based geometric regularisation to align irregular fragment supports with the ViT patch grid, and a supervised contrastive objective that operates on predictive distributions through a Kullback-Leibler similarity and consistently improves every branch. We combine the branches with a deliberately simple learnable logit ensemble. Experiments on CLEOPATRA and POMPAAF show that fragment-aware modelling improves over the standard ViT baseline, with the ensemble increasing accuracy from 0.604 to 0.656 and macro-F1 from 0.596 to 0.648 on CLEOPATRA, and outperforming the best single branch in four of six fragmentation settings on POMPAAF. We additionally evaluate a more complex graph-fusion variant and find that it matches the simple ensemble on POMPAAF while offering only a small, dataset-specific gain on CLEOPATRA, which does not justify its added complexity. Beyond these empirical gains, our contribution is twofold: a distribution-level contrastive objective that consistently sharpens single-branch recognition, and an interpretability analysis that verifies the models exploit genuine painted evidence, while quantifying that the inpainting-based branch draws part of its attribution from the synthesised surround. --- ### On the Limits of Maximal Coding Rate Reduction for Out-of-Distribution Generalisation Authors: Menghui Zhou, Gaoshan Bi, Vitaveska Lanfranchi, Po Yang Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.21001v1 Substantial efforts have been devoted to making deep learning objectives, representations, and architectures interpretable, with the goal of improving the safety, robustness, and generalisation of learning systems in diverse real-world applications. The recently proposed maximal coding rate reduction ($\mathrm{MCR}^{2}$) offers a promising information-theoretic framework for learning structured, discriminative representations of class-wise submanifolds and has inspired interpretable white-box architectures. However, we observe that $\mathrm{MCR}^{2}$ can completely fail under distribution shift, motivating our study of its out-of-distribution (OOD) generalisation limits. We establish two limitations of $\mathrm{MCR}^{2}$ for OOD generalisation. First, the $\mathrm{MCR}^{2}$ objective alone can admit complete prediction failure: a representation based entirely on unstable environmental features can achieve the global coding optimum yet fail completely after correlation reversal, despite an available perfectly stable feature. This exact-optimum example includes test inputs that cannot occur during training. Even when every possible test input can also occur during training, coding quality can be arbitrarily close to optimal while prediction error is arbitrarily close to 100%. Second, directly incorporating the invariance principle underlying widely successful invariant risk minimisation (IRM) and risk extrapolation (REx) does not eliminate this failure. The failing representation admits the same optimal coding operator across training environments, showing that shared coding optimality does not ensure stable prediction. Reliable OOD guarantees for $\mathrm{MCR}^{2}$ therefore require additional new assumptions or learning principles that establish stable predictive relationships across environments. --- ### Do Spinning Radar Doppler Velocity Measurements Improve Vehicle Detection and Tracking? Authors: Eric Xie, Daniil Lisus, Timothy D. Barfoot Published: 2026-09-17 Categories: cs.RO, cs.CV, eess.SP arXiv: https://arxiv.org/abs/2609.21000v1 Spinning frequency-modulated continuous-wave (FMCW) radars have been gaining popularity in autonomous vehicle perception on account of their robustness to adverse weather conditions and 360° field of view. Recently, scanning radars have also been shown capable of generating per-azimuth Doppler velocity. In this paper, we investigate whether these Doppler velocity measurements improve spinning radar vehicle detection and tracking performance. For detection, we estimate the ego motion and use it to undo the Doppler range distortion of the radar image before passing it to a network. For tracking, we propose a new way to estimate a per-vehicle velocity and use it as a prior for the tracker's motion model. Since Doppler-enabled spinning radar data is not available in any dataset with ground-truth dynamic object labels, our first contribution is an automatic labelling pipeline that uses an ensemble of fine-tuned off-the-shelf lidar detectors to label all 643 km of the Boreas Road Trip dataset. We then transfer detections to radar, and use over 250 km of vehicle-dense sequences as ground-truth training data. By training and evaluating two state-of-the-art detectors, we show that Doppler undistortion can improve detection accuracy by up to $2.37$ points on mean average precision. Furthermore, we show that the Doppler velocity prior can improve tracking accuracy by $13.68$ points on multi-object tracking accuracy (MOTA) versus the zero-velocity initialization baseline, while achieving $99.7\%$ of the MOTA obtained using ground-truth velocities as the prior. --- ### Aggregated Posterior Predictive Checks for Generative Modeling Authors: Shweta Dutta, Gemma E. Moran Published: 2026-09-17 Categories: stat.ME, cs.LG, math.ST, stat.ML arXiv: https://arxiv.org/abs/2609.20999v1 Latent variable generative models are commonly fit using simple priors over latent variables, but draws from these priors often fail to produce realistic data. This failure is due to a mismatch between the prior and the aggregated posterior, the distribution of latent variables induced by the fitted model and the data. This mismatch is often viewed as evidence that the prior is misspecified and should be replaced. Alternatively, in modern generative models, a two-stage strategy is increasingly used where first, the model is fit, and second, the aggregated posterior is estimated (van den Oord et al.,2017; Rombach et al., 2022.). Synthetic data are then obtained by sampling from this aggregated posterior instead of the prior. To check such procedures, we introduce the aggregated posterior predictive check (APPC). Theoretically, we establish sufficient conditions under which the APPC is asymptotically calibrated. For probabilistic principal component analysis, we show that the APPC can remain calibrated under a misspecified latent prior when pervasive factors permit recovery of the signal space. Experiments with variational autoencoders show that aggregated posterior sampling improves generation for heavy-tailed and clustered data relative to Gaussian prior sampling while performing comparably to models with more flexible latent priors. --- ### MOSAIC-SR: Transformer-Guided Symbolic Regression for Scientific Equation Recovery Authors: Peiyi Zheng, Yanming Kang, Hans De Sterck, Giang Tran Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20997v1 Symbolic regression aims to recover closed-form equations from observations, providing interpretable models for scientific discovery. Existing approaches struggle to combine flexible structural search with efficient inference. Search-based methods can refine expression structure but often rely on costly combinatorial optimization with random initialization. Pretrained neural models generate formulas almost instantly, but their predictions often contain symbolic errors. We introduce MOSAIC-SR, which uses a pretrained Transformer to propose multiple initial sketches. These sketches initialize searches in several promising regions, avoiding random starts in the vast expression space. Each search jointly recovers structure and constants through scale-aware constant optimization and local symbolic repair. We evaluate MOSAIC-SR on the SRSD-Feynman dataset with and without dummy variables and on six additional benchmarks. MOSAIC-SR obtains the highest symbolic solution rate on every dataset while ranking among the top two methods in predictive accuracy. This advantage persists in the presence of irrelevant dummy inputs. The results show that learned priors can focus search on promising equation structures, and that numerical optimization and symbolic repair are important for recovery. --- ### Voice-Light: A Full-Duplex Cascaded Voice Agent with Causal Turn-Taking and Speculative Generation Authors: Bertil Braun Published: 2026-09-17 Categories: cs.SD, cs.CL, eess.AS arXiv: https://arxiv.org/abs/2609.20995v1 Natural spoken interaction requires more than streaming ASR, language generation, and speech synthesis: a system must react to overlap without canceling on every acknowledgment, prepare a response before a turn is certain, and ensure canceled audio cannot enter conversation history. We present Voice-Light, a full-duplex cascaded voice agent that combines immediate acoustic onset, a causal adapter sharing a streaming ASR encoder, reversible playback control, and private speculative response generation. Structured tool calls execute concurrently with audible bridge speech, while browser acknowledgments make rendered audio authoritative for durable history. Locked evaluation on 1,673 real-conversation silence candidates found that an earlier learned completion checkpoint preserved a 2.70% false-cutoff rate but reached only 12.53% end-of-turn recall, compared with 95.60% for a Silero timing policy. The deployed system therefore retains a hybrid controller rather than claiming a learned-policy replacement. Across three unscripted operator-run microphone sessions, 36 measured response turns had a 758 ms median from final VAD endpoint to first server audio; 21 turns were below 800 ms. These sessions are an instrumented case study, not a controlled user evaluation. We release the synthetic data, model artifacts, evaluation code and summaries, source code, and deployment configuration supporting the result. --- ### From Stress to Affect: Multimodal Deep Learning for Physiological Emotion Recognition Across Wearable Sensor Modalities Authors: Desta Haileselassie Hagos, Saurav Keshari Aryal, Legand L. Burge Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20991v1 Physiological emotion recognition using wearable sensors has important applications in mental health monitoring, affective computing, and human-computer interaction. However, existing studies typically evaluate a single model, sensing configuration, or dataset, limiting our understanding of how these factors influence recognition performance. We present a comparative study of temporal deep learning architectures for physiological emotion recognition using two multimodal wearable datasets: WESAD and EmoWear. Bidirectional long short-term memory (LSTM), temporal convolutional network (TCN), and Transformer models are evaluated under wrist-only, chest-only, and multimodal sensing configurations using participant-independent leave-one-subject-out cross-validation (LOSO-CV). We also investigate soft-voting ensembles, sensor ablation, sampling frequency, and gradient-based saliency. The Transformer achieved the highest multimodal accuracy on WESAD (99.02% +/- 0.51%), whereas the LSTM achieved the best multimodal accuracy on EmoWear for both arousal (91.80% +/- 1.06%) and valence (89.96% +/- 0.36%). These results show that relative architecture performance depends on dataset characteristics rather than one architecture being uniformly superior. Multimodal sensing consistently outperformed wrist-only and chest-only configurations across both datasets. Sampling-frequency analysis showed that 4 Hz provides a practical operating point, with performance comparable to higher frequencies at substantially lower training cost. These findings provide guidance for selecting architectures, sensing modalities, and sampling frequencies for wearable physiological emotion recognition. --- ### Trustworthy FinAInce: Unpacking How AI-Mediated Financial Advice is Judged Authors: Aryan Ramchandra Kapadia, Eshwar Chandrasekharan, Koustuv Saha Published: 2026-09-17 Categories: cs.HC, cs.AI, cs.CL, cs.CY arXiv: https://arxiv.org/abs/2609.20989v2 As generative AI is increasingly used as a source of personal financial guidance, understanding how people appraise such advice is important for supporting appropriate reliance. We conducted a randomized vignette experiment with 285 U.S. adults across eight financial decisions, independently varying three advice styles---AI, expert, and online community---and displayed source labels while holding the underlying recommendation consistent. Advice style most strongly shaped message and safety appraisals, Expert labels selectively increased perceived source knowledge, and decision context primarily shaped risk and safety appraisals. These appraisals were associated with downstream judgments, with models explaining 69.2% of overall quality, 75.9% of trust, and 82.9% of intended reliance. Expert-style advice also remained most preferred when shown without source labels. Our findings have implications for understanding financial advice evaluation, distinguishing the roles of advice style and source labels, and designing financial AI that supports grounded evaluation rather than simply maximizing trust. --- ### ASGARD: Action-Space Guard for UAV Resilience via Reinforcement Learning Authors: Mohsen Salehi, Karthik Pattabiraman Published: 2026-09-17 Categories: cs.LG, cs.CR, cs.RO arXiv: https://arxiv.org/abs/2609.20982v1 Reinforcement learning (RL) controllers have been recently adopted for Unmanned Aerial Vehicles (UAV) navigation and control. However, they are susceptible to action-space attacks that overwrite the action commands after the policy generates them and before the actuators execute them. While most existing defenses target attacks on the policy's inputs, those addressing action-space attacks retrain the policy at training time and are not resilient to corrupted actions at runtime. We propose ASGARD, a two-phase teacher-student pipeline for making RL-based UAV control resilient to action-space attacks. In the teacher phase, an encoder combines the UAV's physical state with action-attack-related privileged information to produce an action-attack-aware latent that trains the RL control policy and a monitor that outputs corrected action commands to the actuators. In the student phase, both the encoder and the monitor are trained via supervised learning from their teacher counterparts to run on-board using only the UAV's physical state history. We evaluate ASGARD across attack scenarios targeting different action commands on UAV. We find that ASGARD is resilient to action-space attacks and completes the missions despite the attack. We further find that ASGARD generalizes to unseen attacks and remains resilient against stealthy attacks. --- ### CaLR: Causal Latent Revision for Robust Diffusion Reasoning Authors: Wei Cai, Jian Zhao, Yuchen Yuan, Xuelong Li Published: 2026-09-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.20981v1 Autoregressive (AR) models suffer from local greediness, while diffusion language models (DLMs) often lack the strict causal structure required for reasoning. To combine the advantages and overcome the drawbacks of the dual, we propose Causal Latent Revision (CaLR), a framework that reformulates reasoning as constrained latent optimization. By adopting a causal topology matrix (CTM) from an expert model and implicit differentiation, CaLR performs gradient-guided ``thought revision" to enforce logical consistency, enabling dynamic self-correction of intermediate steps during parallel generation. Empirically, CaLR achieves SOTA DLM performance on complex benchmarks, surpassing strong AR baselines and demonstrating superior robustness in constrained tasks like Sudoku. --- ### Ranking Competing geologic interpretations via foundation-model-assisted generative hydrologic inversion Authors: Harun Ur Rashid, Daniel O'Malley Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20978v2 High-consequence subsurface decisions often rely on sparse data that permit competing geological interpretations. Determining consistency of these interpretations with the available observations remains challenging. We present a workflow that addresses this challenge by translating competing geologic interpretations into alternative priors and ranking them according to their consistency with hydraulic-head observations. A key step in this workflow is exploiting the broad knowledge of image-generation foundation models to transform nuanced geologic interpretations into data ready for computer modeling. For each interpretation, a text-to-image foundation model generates an ensemble of geologic images, and a separately trained variational autoencoder learns an interpretation-specific latent representation. A supervised inverse network maps head observations into this latent space, and the frozen decoder reconstructs an image that is mapped to a log-conductivity field. Steady-state flow simulations predict heads, and the aggregate normalized head error determines the ranking. We evaluate the framework using a synthetic benchmark based on the Johansen Formation with three interpretations of decreasing consistency with the reference geology. Across 595 test cases, the Precise \& Accurate interpretation produces lower normalized errors than Accurate in 58.5\% of cases and Mismatched in 82.5\% of cases. Accurate outperforms Mismatched in 65.5\% of cases. We then compare spatial representations of two published conceptual models of the Culebra Dolomite Member at the Waste Isolation Pilot Plant. The revised representation yields an aggregate normalized error of 7.598, compared with 8.595 for the original, consistent with the documented conceptual-model revision. The framework enables quantitative comparison of competing geological interpretations using available hydraulic observations. --- ### Image-Derived PM10 Estimation in Cattle Feedlot Using Machine Learning: Addressing Concentration Ranges Beyond Existing Digital Imaging Methods Authors: Sirapoom Peanusaha, Greg B. Ferguson, K. Jack Bush, Peiyang Li, Brent W. Auvermann Published: 2026-09-17 Categories: cs.CV, physics.optics arXiv: https://arxiv.org/abs/2609.20975v1 Affordable dust monitoring remains a pressing need for the cattle feedlot industry, yet camera-based PM estimation, despite its growing body of research in urban air quality settings, has not been evaluated under the extended concentration ranges characteristic of intensive livestock operations. This study developed an image-based approach using contrast panel features and machine learning to estimate PM10 concentrations in a commercial cattle feedlot, where hourly average PM10 ranged from 250 to 1,000 ug/m^-3 and instantaneous concentrations reached 5,000 to 20,000 ug/m^-3. Grayscale images were captured during the evening dust peak period, and features including panel contrast, black and white panel pixel values, and overall image brightness were extracted. The model also incorporated recent past values from preceding images and solar zenith angle as predictors. Among the candidate models evaluated, XGBoost achieved the highest predictive performance, with an R^2 of 0.792 and a median absolute error of 103 ug/m^-3. Feature importance analysis revealed that (a) panels positioned farthest from the camera contributed most strongly to predictions and (b) that black panel pixel values were more sensitive than white panel values to changes in PM10 concentration. Prediction accuracy during the sunset transition, which coincides with the onset of the feedlot evening dust peak, remains an area for further refinement. These findings demonstrate the feasibility of image-based PM10 estimation across PM concentration ranges substantially exceeding those reported in prior urban studies and provide practical guidelines for future deployment in feedlot environments. --- ### Attention-Aware Routing: Coupling Routing and Attention in MoEs Authors: Despoina Kosmopoulou, Anastasios Tsetsilas, Efthymios Georgiou, Giannis Karamanolakis, Swastik Roy, Alexandros Potamianos Published: 2026-09-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.20974v1 In Mixture-of-Experts language models, the router typically selects and weights experts based on the token's hidden state, utilizing limited contextual information. We propose Attention-Aware Routing (AAR), which augments the router with temporal and spectral features extracted from a sliding window of attention weights that represent a summary of the model's contextual state, disentangled from the hidden state. Keeping the base transformer entirely frozen, we train only the routing parameters, isolating routing as the sole variable. AAR improves GSM8K by +3.37 pp over a routing-only SFT baseline on OLMoE. Beyond performance, we show that routing and attention form a coupled circuit: routing changes at layer l propagate through the residual stream to amplify attention sinks at layer l+1, reshaping attention without any direct update to the attention mechanism itself. Further, AAR reduces long diverging generation, with incorrect answers getting shorter, while correct answers remain unchanged in length. Finally, AAR is strongly depth-sensitive: applying it indiscriminately across layers can degrade factual retrieval, whereas mathematical reasoning gains persist when it is introduced deeper in the network. This sensitivity exposes a retrieval--reasoning tension across depth and makes layer-selective AAR a controlled probe of the routing-relevant information carried by attention at different layers. --- ### Complex Problem Solving in Large Language Models: A Statistical Control Survey and Diagnostic Framework Authors: Jiazhang Cai, Tao Wang, Ruidong Zhang, Siyuan Li, Terry Ma, Luyang Fang, Haoran Lu, Huimin Cheng, Yingchuan Zhang, Shushan Wu, Rui Xie, Lin Tang et al. Published: 2026-09-17 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2609.20973v1 Complex problem solving (CPS) with large language models (LLMs) is often framed as a matter of stronger reasoning or longer generation. Yet early-step error amplification, prompt brittleness, and failures to revise incorrect commitments are difficult to explain by missing knowledge or expressive capacity alone. This survey interprets CPS as a sequential estimation-and-decision problem over a latent solution state. A controller maintains a belief about an unobserved solution trajectory, updates it as noisy intermediate evidence arrives, and decides whether to commit, verify, branch, roll back, or abstain to minimize expected loss. Reasoning supplies candidate transitions and interpretations, whereas process control shapes and evaluates those proposals and regulates subsequent transitions and observations. Within this framework, we organize existing methods around five components: explicit state representation, transition structuring, validation and constraint enforcement, search and rollback, and uncertainty management. We also interpret evaluation metrics according to the statistical quantities they estimate. The framework further yields a diagnostic hypothesis: interventions should be most effective when they target the error or uncertainty component implicated by an observed failure. We distinguish systematic, stochastic, and irreducible error together with epistemic and aleatoric uncertainty, and call this alignment problem-control fit and its failure control mismatch. For example, additional sampling may reduce sampling variability while leaving a shared systematic error unchanged. This perspective clarifies what current methods estimate and control, what remains uncontrolled, and why reliable validation, targeted recovery, calibrated uncertainty, and matched-budget evaluation are central open problems. --- ### RBS-Attention: Radius-Bounded Sparse Prefill for Long-Context Large Language Models Authors: Chuxu Song, Jiuqi Wei, Zhencan Peng Published: 2026-09-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.20971v1 Long-context large language model inference is increasingly limited by prefill, where dense self-attention processes the entire prompt before generation begins. Sparse block selection can reduce this cost, but a block centroid may hide a highly relevant token among many irrelevant ones. We call this failure mode mean dilution and propose RBS-Attention, a training-free sparse-prefill method with two complementary selection branches. A centroid base branch captures average relevance, while a rescue branch uses the maximum key-block radius and its prompt-, layer-, and head-dependent distribution to identify blocks at risk of underestimation. Independently thresholding the two branches and combining their masks controls the contribution of rescue blocks while preserving regular block-sparse FlashAttention execution. On H100 GPUs, RBS-Attention achieves 20.65$\times$ standalone prefill-attention speedup, 11.92$\times$ vLLM prefill-attention speedup, and 5.97$\times$ end-to-end time-to-first-token speedup at 128K on Qwen3-30B-A3B-Instruct-2507-FP8. On the dense Qwen3-32B model, it obtains 88.65 overall RULER accuracy versus 89.52 for dense attention; LongBench-v2, InfiniteBench, and Video-MME provide additional quality evaluation. Supporting experiments measure actual retention, compare selectors at matched density, and characterize block-size, threshold, and memory behavior. Together, these results support radius-adaptive dual-branch selection as an effective approach to long-context prefill. --- ### From Switching to Dynamic Regret: A Simple Reduction via Unbiased Random Sequences Authors: Yibo Wang, Wenhao Yang, Sifan Yang, Yuanyu Wan, Lijun Zhang Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20968v1 In non-stationary online learning, dynamic regret has attracted increasing attention as a measure of how well an online learner performs against a time-varying comparator sequence. Despite considerable advances, attaining optimal bounds for strongly convex and exp-concave losses often involves intricate analysis. In this paper, we present a \textit{simple} framework that reduces dynamic regret minimization to switching regret minimization. As a result, we can derive dynamic regret bounds by using off-the-shelf algorithms with switching regret guarantees. The key idea of our reduction is to construct, for \textit{any} comparator sequence, an auxiliary random sequence that is unbiased at each round, with the controlled variance and a manageable number of switches. Combining this construction with suitable surrogate losses, we can decompose dynamic regret into the expected switching regret against the random sequence and its controlled variance. Theoretically, for strongly convex and exp-concave losses, we establish the $\widetilde{O}(T^{1/3}P_T^{2/3})$ dynamic regret bounds, where $T$ denotes the time horizon and $P_T$ denotes the path-length of the comparator sequence. Moreover, for general convex losses, the same reduction also recovers the $O(\sqrt{T(1+P_T)})$ dynamic regret bound. Notably, all our findings match the minimax optimal results for these three types of losses, highlighting the versatility of our proposed framework. --- ### MemeTAG: Keyword-Driven Meme Classification through Tag Embedding Reconstruction Authors: Akshit Sharma, Prashant W. Patil Published: 2026-09-17 Categories: cs.CV, cs.MM arXiv: https://arxiv.org/abs/2609.20962v1 The proliferation of harmful internet memes poses a significant societal threat, yet their automated classification remains a formidable algorithmic challenge due to the nuanced, multimodal nature of their content. To address this, we introduce MemeTAG, a novel dual-objective framework that pioneers a keyword-aware approach to meme classification. Our core innovation is a two-part semantic guidance mechanism: first, we leverage a pretrained Vision-Language Model to generate a set of descriptive keywords, that capture the high-level semantics. Second, we introduce the Aggregated Tag Inference Network (ATIN), an attention-based module that distills these keywords into a single, rich semantic embedding. This embedding serves as a target for a novel auxiliary reconstruction loss, which compels the model to learn deeply aligned visual and textual features. This approach, combined with an efficient three-stage training strategy, establishes a new state-of-the-art on the HarMeme, Hateful Memes Challenge (HMC), and PrideMM datasets, decisively outperforming existing state-of-the-art methods. --- ### Efficient Bayes-Adaptive Reinforcement Learning with Temporal Logic Specifications Authors: Jonathan Hau, Alessandro Abate Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20954v2 We present a novel end-to-end model-based Reinforcement Learning (RL) algorithm for efficient policy synthesis under given Linear Temporal Logic (LTL) specifications (e.g., safety or reachability) in unknown environments. To do so, a Limit-Deterministic B{ü}chi Automaton (LDBA) representation of the LTL task is synchronised with a Bayes-Adaptive Markov Decision Process (BAMDP) representation of the environment, which allows us to leverage an enhanced exploration-exploitation trade-off that is achieved via Bayesian RL, as opposed to traditional non-Bayesian approaches. We further propose a novel Bayes-Adaptive Monte-Carlo Planning (BAMCP) algorithm to allow for approximate Bayes-optimal strategy synthesis in the synchronised BAMDP construct. A range of finite- and infinite-horizon task experiments demonstrate the effectiveness of our approach in terms of both property satisfaction and sample efficiency, when compared to traditional model-free approaches. Additional ablation studies also successfully highlight the value of the novel BAMCP algorithm in comparison to classical BAMCP for LTL task satisfaction. Finally, we also showcase a successful application of our approach for \textit{cautious} RL, namely to reduce the number of task violations incurred during policy training. --- ### $μ^2$-Bench: A Multilingual Machine Unlearning Benchmark Authors: Kyomin Hwang, Hyeonjin Kim, Hyunho Lee, Yearim Kim, Yeji Song, Nojun Kwak Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20945v1 Undesired information such as harmful content and private data propagates through Multilingual Large Language Models (LLMs) via direct training and indirect cross-linguistic spread. Multilingual Machine Unlearning (MMU) aims to remove such information, yet its evaluation remains underexplored, leaving unclear whether unlearning truly eliminates target knowledge across all languages. To bridge this gap, we introduce $μ^2$-Bench, an MMU benchmark that simulates the full pipeline of memorization, unlearning, and evaluation across diverse languages. It 1) spans a broad set of languages, 2) evaluates on both training and hold-out languages, and 3) assesses knowledge as dispersed across multiple languages. We show that successful MMU requires methods that reflect multilingual characteristics, and conduct analysis to provide deeper insights into MMU. --- ### When AI Reviews Train AI Reviewers: Scientific-Judgment Collapse and Mitigation Authors: Sy-Tuyen Ho, Minghui Liu, Furong Huang Published: 2026-09-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20942v1 Large language models (LLMs) increasingly participate in scientific evaluation, both as automated reviewers and as assistants to human reviewers. As model-generated reviews enter public data and future training corpora, AI peer review can become recursive: later reviewers learn from judgments produced by earlier models. We study one step of this feedback loop in a controlled setting. Starting from Llama 3.1 8B, we first fine-tune a reviewer on official ICLR reviews from 2018--2023 and then train four successor models on ICLR 2024 data with systematically varied mixtures of official and model-generated reviews. Our study shows that introducing synthetic reviews compresses rating distributions and reduces both same-paper and corpus-level semantic diversity. We call this pattern $\textbf{scientific-judgment collapse}$. To mitigate this failure mode, we introduce $\textbf{TrustReviewer}$, an open-source LLM-based system for generating peer reviews of AI and machine learning papers. TrustReviewer intervenes at two complementary stages. For training-time prevention, we train the core reviewer in a single stage on a curated corpus designed to reduce low-quality and semantically degenerate supervision. For test-time correction, paired activation steering aims to further mitigate residual tendencies toward collapsed judgments without further training or additional expert annotation. Together, these results characterize a concrete risk of recursive reviewer training and provide practical interventions for preserving judgment diversity and improving recommendation alignment in AI-assisted scientific evaluation. --- ### Do Quantum Models Scale Like LLMs? Authors: David S. Berman, Ying-Jer Kao, Roger G. Melko, Alexander G. Stapleton Published: 2026-09-17 Categories: cs.LG, quant-ph arXiv: https://arxiv.org/abs/2609.20912v1 In this work, we study the neural scaling laws of RydbergGPT, an autoregressive transformer model trained on qubit projective measurement data gathered from interacting Rydberg atom arrays. The quantum system is known to exhibit a finite-size remnant of a critical point as the laser detuning parameter is varied. We find that near the critical point the transformer loss as a function of training dataset size is well described by a power-law with a loss floor correction. However, away from criticality the quality of the power-law description is substantially reduced. We then compare the statistical structure of both Rydberg measurements and natural-language corpora using an entropy-normalised, finite sample corrected mutual information "two-point" function. We find that near-critical statistics of the two point functions are closest to those observed in natural-language, whilst other qubit configurations far from the critical point have two-point functions that decay more rapidly. This supports the hypothesis that multi-scale dependence contributes to stable neural scaling, and that scaling behaviour should be viewed as a property of the model-data pair. --- ### Continuous Delayed-Memory Stochastic Gradient Descent and Continuous-Time Reinforcement Learning from History of Astrophysical Time Series Studies Authors: Juncheng Yi Published: 2026-09-17 Categories: cs.LG, stat.AP arXiv: https://arxiv.org/abs/2609.20906v2 Quasars are luminous objects in the universe that exhibit stochastic brightness variations encoding information about the supermassive black holes powering them, and modeling these variations from ground-based survey data time series, known as light curves, is a statistical challenge. This paper reviews how stochastic differential equations (SDEs) have been adapted with neural network parameterizations to overcome this challenge in history. We create the Continuous-Delayed-Memory Stochastic Gradient Descent which depend on the past state of the discrete iteration process. We performed the simulation on some 2-dimensional landscape and observed some wider-exploration and more precise convergent behavior compared to Vanilla SGD by adjusting hyperparameters. Besides, we proposed a reinforcement learning structure with continuous time policy gradients for exploratory policies without solving HJB PDE, and we show that its optimality conditions recover the Gibbs policy of previous works. --- ### Uncertainty-driven training for three-dimensional calibrated lung nodule classification Authors: Giuseppe Tripodi, Alessandro De Rosis, Saleh Rezaeiravesh Published: 2026-09-17 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2609.20905v1 In this work, we present an uncertainty-driven training framework for three-dimensional computed tomography (CT) lung nodule classification, where validation-based uncertainty estimates guide loss reweighting to enhance predictive performance and probability calibration. Two Uncertainty Quantification (UQ) methods are considered: Monte Carlo Dropout (MCD) and Evidential Deep Learning (EDL). Both provide per-class uncertainty estimates that modulate the loss and encourage focus on hard or unreliable classes. The framework is evaluated with ResNet, DenseNet, EfficientNet, Vision Transformer (ViT), and Swin Transformer backbones on two datasets: the clinical LIDC-IDRI cohort and the NoduleMNIST3D benchmark. Uncertainty-driven training achieves classification performance similar to conventional training while substantially improving calibration, with an expected calibration error (ECE) reduced by up to 65% on LIDC-IDRI. EDL attains competitive performance on shallower architectures with single-pass inference, whereas MCD is more robust on deeper networks. Analysis across architectural families reveals that uncertainty-driven training benefits convolutional backbones more consistently than transformer-based architectures: EDL in particular degrades on ViT, suggesting that the Dirichlet evidence parameterisation may interact unfavourably with attention-based architectures at lower input resolutions. A posteriori temperature scaling proves highly effective across all configurations, indicating that a simple scalar calibration can be competitive even without explicit uncertainty-aware training. Our results indicate that integrating UQ into the training loop can significantly improve probabilistic calibration and support more trustworthy deployment of three-dimensional medical imaging models. --- ### Bio-MF: Low-Latency and High-Fidelity EEG-to-fNIRS Cross-Modal Generation for Hybrid Motor-Imagery Brain--Computer Interfaces Authors: Boyuan Zhao, Sifan Zhang, Luping Chen Published: 2026-09-17 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.20904v1 Hybrid motor-imagery brain-computer interfaces (MI-BCIs) combining EEG and fNIRS can outperform EEG-only systems by exploiting complementary electrophysiological and hemodynamic information. To obtain such hybrid information when paired EEG-fNIRS acquisition is unavailable or inconvenient, recent studies have focused on EEG-to-fNIRS cross-modal generation. However, existing methods still suffer from slow generation and often require pretraining, limiting their use in real-time MI-BCI scenarios. Although one-step generative models offer an attractive route to low-latency synthesis, removing the iterative refinement process can reduce generation fidelity and introduce non-physiological artifacts. To address these problems, this paper proposes Bio-MF, a latent-free one-step MeanFlow framework for EEG-conditioned fNIRS generation. Bio-MF performs direct signal-space x-prediction, converts this signal-space output into MeanFlow velocity supervision, and completes inference with one network evaluation. To preserve task-relevant hemodynamic structure under heterogeneous sensor layouts, Bio-MF integrates Spatial-Temporal Interactive 4D Encoding, cross-modal classifier-free guidance, and noise-level-gated FFT regularization. On Dataset 1, EEG + synthetic fNIRS improves ACC over EEG-only by 3.37 and 4.15 percentage points for HbR and HbO, respectively. On Dataset 2, the corresponding gains remain 2.98 and 2.50 percentage points under the unseen 64-channel EEG montage. On an RTX PRO 6000 GPU, Bio-MF generates one fNIRS trial in 7.0 ms, corresponding to an 857x speedup over the 1000-step SCDM latency. These results show that Bio-MF enables fast EEG-to-fNIRS synthesis while preserving task-relevant generation quality for downstream hybrid MI decoding. Our code is available at https://github.com/psychosiwa/Bio-MF. --- ### Generative Artificial Intelligence Chatbots for Motivational Interviewing: A Scoping Review From System Design to Intervention Outcomes Authors: Runze Hu, Jingqi Kong, Yang Yang, Yihang Yang, Jingyao Liu, Haizhou Tang, Shanghang Zhang, Zheng Liu Published: 2026-09-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20902v1 Motivational interviewing (MI) is a collaborative approach to elicit autonomous motivation for health behavior change. Generative AI (GenAI) offers new ways to deliver MI via conversational systems, but evidence on their design, assessment, and translation into interventions remains fragmented. This scoping review characterized evidence on GenAI-MI chatbots across system design, safety, MI quality, user perceptions, and intervention outcomes. We conducted a PRISMA-ScR scoping review. Nine datasets were searched for studies published or publicly available from January 1, 2015 to June 2, 2026 that used GenAI to generate MI chatbot responses or counselor utterances. Data were extracted using a predefined framework and synthesized descriptively. Forty-seven reports (48 studies) were included. Twenty (41.7%) focused on system design without direct participant use; 28 (58.3%) involved direct interaction. Most systems were text based and disembodied; 23 (47.9%) incorporated dynamic adaptation. Safety measures were unevenly reported. Among studies with direct use, 21/28 (75.0%) reported informed consent or user education. Thirty (62.5%) assessed MI quality, generally suggesting MI-consistent interactions. User perceptions were favorable, especially empathy, usability, helpfulness, and intention to use, though measures were heterogeneous. Eighteen (37.5%) reported intervention outcomes, mostly after a single session. Positive findings were more consistent for short-term motivation than sustained behavioral or functional change. GenAI-MI chatbots can deliver MI-consistent interactions perceived favorably, but evidence for sustained behavioral or functional change is limited. Future research should strengthen runtime safety monitoring, standardize MI quality assessment, and use longer-term comparative designs with behavioral and functional outcomes. --- ### SpaceDiffusion: Over-the-Orbit Diffusion for Space Generate-and-Forward Communications Authors: Jianhao Huang, Zhanwei Wang, Khaled B. Letaief, Kaibin Huang Published: 2026-09-17 Categories: cs.IT, cs.AI arXiv: https://arxiv.org/abs/2609.20899v1 Satellite communications are an essential component of sixth-generation (6G) mobile networks, which provide ubiquitous connectivity for global services. However, the satellite uplink remains a critical bottleneck for ground devices: their limited transmit power and antenna apertures result in low data rates and high packet errors. To overcome this bottleneck, this paper advocates a novel relaying paradigm termed generate-and-forward (GF) communications, where satellites exploit on-orbit generative artificial intelligence (AI) to robustly reconstruct corrupted data prior to forwarding. Specifically, we propose SpaceDiffusion, an over-the-orbit diffusion framework for satellite-assisted image transmission. The core of this framework is a channel-distortion-aware diffusion theory developed using the following approach. By formulating the recovery of compressed and lost image tokens as an inverse problem, this theory incorporates a channel-distortion correction term directly into the conventional denoising diffusion implicit model (DDIM) update. As a result, this design enables a single pretrained diffusion model to adapt dynamically to varying packet-loss patterns and compression distortions without retraining. Furthermore, we analytically characterize the progressive token-reconstruction error and derive a diffusion-step activation threshold that predicts when SpaceDiffusion is expected to outperform conventional decode-and-forward (DF) relaying. Building on these theoretical insights, we further develop an energy-aware early-exit policy to efficiently deploy SpaceDiffusion in orbit. Experimental results demonstrate that SpaceDiffusion achieves lower end-to-end latency compared to DF scheme with retransmission protocol and saves approximately 15 dB of uplink transmit power at a target perceptual quality. --- ### Extreme classification: beating chance with one training example from each class Authors: Kevin Bleakley, Aaditya Ramdas Published: 2026-09-17 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2609.20897v1 We study a minimal classification problem: Given independent labeled observations $X\sim P$ and $Z\sim Q$ from two unknown distributions $P,Q$, and given an independent target $Y$ drawn with equal probability from $P$ or $Q$, can one classify $Y$ strictly better than chance whenever $P\neq Q$? The one-nearest-neighbor rule succeeds for every pair of multivariate Gaussian distributions with distinct means and a common positive-definite covariance matrix but can perform strictly worse than chance even for smooth densities on the real line. We construct a fixed randomized kernel rule whose expected accuracy is exactly $1/2+\operatorname{MMD}_k^2(P,Q)/4$, and obtain characteristic kernels on countably generated measurable spaces from countable families of measurable binary questions. We also prove that a deterministic order rule on $\mathbb R$ beats chance for every pair of distinct Borel probability measures. A measurable encoding then gives a deterministic distribution-free rule which beats chance on every countably generated measurable space, in particular every separable metric space. Finally, we show that no rule works for every distinct pair of distributions and every unknown unbalanced class prior; under adaptive target-class selection, every rule other than a fair coin is strictly worse than chance for some finitely supported pair. --- ### WM-VS: Progress-Aligned World Models for Closed-Loop Visual Servoing Authors: Guanzhong Sun, Junyi Ma, Yixuan Zhou, Yuxuan Wu, Yanzi Miao, Hesheng Wang Published: 2026-09-17 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2609.20892v1 Closed-loop visual servoing requires predictions that indicate whether an action reduces task error, not only whether the action is plausible. We call this gap the prediction-control mismatch and introduce WM-VS, a target-centric progress-aligned world-model framework for closed-loop visual servoing. Offline target-region DINOv2 correspondences define a signed four-dimensional servo coordinate for translation, scale, and in-plane rotation. Stage 1 aligns action-conditioned latent transitions with this coordinate; Stage 2 freezes the world model and trains a reactive joint-velocity policy with action imitation, consequence supervision, and short imagined rollouts that favor error contraction. Deployment is RGB-only and reactive, without online trajectory optimization. On a real 7-DoF eye-to-hand system, WM-VS reaches a corner RMSE no larger than 10 percent of its initial value in 30/30 trials and retains this criterion at the final valid frame in 25/30 (83.33 percent). Removing future-error alignment reduces retention to 26.67 percent. The learned progress signal agrees with an external AprilTag corner error not used for training or control (mean Spearman rho = 0.8778). Without retraining, two unseen 3D targets achieve translation-error reductions of 86.48 percent and 90.27 percent and rotation-error reductions of 70.01 percent and 65.70 percent. These results link progress-aligned action consequences to repeated closed-loop correction and transfer. Code and data will be released as open source. --- ### Coding Agents with an Obstacle-Aware Harness for Safe Robot Manipulation Authors: Bingxin Xu, Yuzhang Shang, Zhen Dong, Emilio Ferrara Published: 2026-09-17 Categories: cs.RO, cs.AI, cs.CL, cs.CV arXiv: https://arxiv.org/abs/2609.20822v1 Coding agents have emerged as a promising paradigm for robot manipulation: a language model writes the robot controller as a program, and agents built in this way now operate robots without robot-specific training.Whether this paradigm is also safe, however, has not been asked. We evaluate coding agent under a safety constraint, where each task pairs a manipulation goal with an obstacle the robot must not touch. The agent pursues the goal but collides with the obstacle in most cases, treating task completion as its sole objective while neglecting safety. The agent reasons about the obstacle in its traces, and the prompt already forbids touching it, so neither perception nor instruction is at fault; the fault lies in the planning, where the stated constraint never becomes a priority. By decomposing manipulation into a route phase and a contact-rich moment, we locate the source of the failure. Along the route, the model cannot prioritize the safety constraint, having no notion of a clearing route and none of replanning once a chosen route becomes infeasible. At the contact, it is unaware that contact execution is bounded by the same constraint. To close this gap, we present SafeHarness, which equips the model with two obstacle-aware harnesses that enable it to prioritize the safety constraint. Obstacle-aware route planning grounds the objects as bounding boxes and draws candidate routes over them as sequences of waypoints. The agent then plans a route in advance, verifies it, replans when necessary, and only then executes it. Obstacle-aware contact execution instead selects the contact position so that the contact itself avoids the obstacle. SafeHarness attains 71.9% task success and 87.5% collision avoidance, surpassing the previous SOTA by 6.5% and 27.0%, respectively. These results are $2.3\times$ and $1.5\times$ those of the same agent without harnesses. --- ### Embedding Models Measure in Peculiar Ways Authors: Juri Opitz, Andrianos Michail Published: 2026-09-17 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.20821v1 Embedding spaces define notions of semantic similarity and distance. We study whether those embeddings reflect physical measurements of mass, distance, time and volume, which admit a unique, objective notion of semantic equivalence and distance. We find that physical measurement is only weakly modeled in the embedding space, and that instead quite peculiar measurement patterns can be observed. Further analysis indicates that embedding representations of physical measurements are strongly influenced by superficial string similarity, and recalibration of similarity does not substantially improve the alignment. --- ### Workspace Models: Lightweight Robotic Memory via Saliency-Driven Supervision Authors: Nitish Dashora, Douglas Chen, Idan Shenfeld, John Marangola, Pulkit Agrawal, Max Simchowitz Published: 2026-09-17 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.20820v2 Complex robotic manipulation tasks frequently require a long-term memory of past events and actions. As conditioning on full histories renders policies prone to spurious correlations and degrades performance, many approaches to policy memory involve compressing historical information through expensive VLM queries in-the-loop to process only task-salient information. In this paper, we propose an alternative approach in which computationally intensive VLM queries are made during train-time to learn a lightweight latent memory that can be efficiently queried at deployment time. Our representation, which we call the \textbf{workspace token}, is trained by (1) using a VLM to identify current and historical information necessary for completing a task, then (2) distilling these into the workspace token using a set-reconstruction decoder loss. In both simulation and hardware, we show that the workspace token can be used as a drop-in replacement for observations during deployment, enabling policies to solve memory-intensive tasks without the need for VLM reasoning in-the-loop, in effect serving as a \textbf{latent harness} for distilling a stronger reasoning models ability to solve long-horizon tasks to a reactive robotic policy. We further demonstrate that the workspace tokens are not only more lightweight, but also lead to better policy performance compared to conditioning policies on explicit modalities like curated past image frames, motivating a ``latent'' approach to history curation and reasoning model harnesses more broadly. --- ### Workspace Models: Lightweight Robotic Memory via Saliency-Driven Supervision Authors: Nitish Dashora, Douglas Chen, Idan Shenfeld, John Marangola, Pulkit Agrawal, Max Simchowitz Published: 2026-09-17 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.20820v1 Complex robotic manipulation tasks frequently require a long-term memory of past events and actions. As conditioning on full histories renders policies prone to spurious correlations and degrades performance, many approaches to policy memory involve compressing historical information through expensive VLM queries in-the-loop to process only task-salient information. In this paper, we propose an alternative approach in which computationally intensive VLM queries are made during train-time to learn a lightweight latent memory that can be efficiently queried at deployment time. Our representation, which we call the \textbf{workspace token}, is trained by (1) using a VLM to identify current and historical information necessary for completing a task, then (2) distilling these into the workspace token using a set-reconstruction decoder loss. In both simulation and hardware, we show that the workspace token can be used as a drop-in replacement for observations during deployment, enabling policies to solve memory-intensive tasks without the need for VLM reasoning in-the-loop. Interestingly, we found that workspace tokens are not only more lightweight but also lead to better policy performance. --- ### Can 4D Foundation Models Remember? Authors: Guangzhao He, Hadar Averbuch-Elor, Wei-Chiu Ma Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20819v2 Perceiving and remembering the visual world is fundamental to navigating and interacting with our environment. Current 4D foundation models, such as camera-controllable video models or 4D reconstruction models, can perceive and reconstruct dynamic environments, but how well they remember what they have perceived remains an open question. Existing benchmarks largely rely on pixel-level metrics and lack ground truth for objects once they leave the field of view, making them unable to evaluate visual memory in an object-centric manner against references. To fill this gap, we introduce PersistBench, a dataset and metric suite that leverages 360° videos as omniscient ground truth and proposes three evaluation aspects: object permanence, motion continuity, and appearance preservation. Evaluating various models across diverse categories reveals that current models can only maintain short-term consistency that degrades significantly once objects leave the field of view. Our findings highlight the gap between current model capabilities and robust visual memory ("seeing is not remembering"), providing guidance for future development of 4D foundation models. Dataset and code are available on the project page: https://guangzhaohe.com/persistbench. --- ### Can 4D Foundation Models Remember? Authors: Guangzhao He, Hadar Averbuch-Elor, Wei-Chiu Ma Published: 2026-09-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20819v1 Perceiving and remembering the visual world is fundamental to navigating and interacting with our environment. Current 4D foundation models, such as camera-controllable video models or 4D reconstruction models, can perceive and reconstruct dynamic environments, but how well they remember what they have perceived remains an open question. Existing benchmarks largely rely on pixel-level metrics and lack ground truth for objects once they leave the field of view, making them unable to evaluate visual memory in an object-centric manner against references. To fill this gap, we introduce PersistBench, a dataset and metric suite that leverages 360° videos as omniscient ground truth and proposes three evaluation aspects: object permanence, motion continuity, and appearance preservation. Evaluating various models across diverse categories reveals that current models can only maintain short-term consistency that degrades significantly once objects leave the field of view. Our findings highlight the gap between current model capabilities and robust visual memory ("seeing is not remembering"), providing guidance for future development of 4D foundation models. Dataset and code are available on the project page: https://guangzhaohe.com/persistbench. --- ### SplashSplat: Reconstructing Splashing Liquids from Real-World Multi-View Videos Authors: Peiyu Liu, Dingxi Zhang, Federico Tombari, Marc Pollefeys, Christina Tsalicoglou, Daniel Barath Published: 2026-09-17 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2609.20818v1 A splash lives for a fraction of a second: sheets tear into ligaments and droplets, appearance is view-dependent and nearly textureless, and little persists long enough to track. Reconstruction research has consequently focused on smoke, synthetic liquids, or gently deforming surfaces. To our knowledge, no synchronized multi-view dataset of splashing liquids exists. We therefore introduce a benchmark of 20 real scenes, from coherent streams to violent splashes, captured by seven synchronized, calibrated 4K cameras at 60 fps, with manually refined per-view liquid and container masks and fixed evaluation splits. We further present SplashSplat, built on a single principle: impose physical structure only where the observations can constrain it. Per-frame liquid SDFs fused from the masks provide the geometry, level-set transport between consecutive SDFs yields a coarse velocity field, and Lagrangian carriers advected along this flow, corrected against each new observation and reseeded where coverage is lost, decode local Gaussians for differentiable rendering. SplashSplat outperforms state-of-the-art dynamic Gaussian splatting methods on our real captures and on a synthetic benchmark, with physically more plausible motion and a lower training cost. The same representation supports temporal interpolation and style transfer without re-optimization. --- ### FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations Authors: Kevin Qu, Tao Sun, Massimiliano Viola, Liyuan Zhu, Zhizhuo Zhou, Sayan Deb Sarkar, Konrad Schindler, Iro Armeni Published: 2026-09-17 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2609.20817v1 Modeling articulated objects from sparse monocular views is challenging because each observation reveals only partial geometry and motion evidence. Most feed-forward methods infer articulation from a single observation and therefore rely heavily on learned category-level shape priors. We present FAMOS, a feed-forward model that predicts movable-part segmentation and joint parameters from a sparse, unordered set of partial point clouds. Our model jointly reasons over multiple observations and naturally supports a variable number of inputs, including a single view. To aggregate articulation cues across observations, we introduce a Multi-state Articulation Transformer with alternating state-wise and global attention. We further propose an observed articulation span objective that supervises the motion range each part exhibits across the input observations, encouraging the model to leverage the full observation set. To overcome the limited scale and diversity of existing datasets, we introduce a procedural data generator that synthesizes self-annotated assets during training. Experiments on PartNet-Mobility, ACD, and ArtiCraft-10K demonstrate consistent improvements over both feed-forward and optimization-based baselines. Project page: https://kevinqu7.github.io/famos --- ### Paint-Anything: Unified Any-Color Control for Image Generation and Editing Authors: Ji Xie, Dewei Zhou, Xinyu Huang, Zhennan Chen, Xun Wang Published: 2026-09-17 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.20816v2 Professional design requires any-color control: the ability to specify an object's target color with any 24-bit hex value for image generation and editing. Prior work has explored color generation, editing, and colorization, but often relies on dedicated color representations or specialized inference procedures. Advances in large language models offer a simpler starting point: even compact models can associate hex values with color semantics. We present Paint-Anything, which learns a shared hex-prompt interface for generation and editing through object-level color supervision. We develop a data pipeline that constructs Paint-500K from real images through object grounding, perceptual color labeling, and editing-pair synthesis. Since shadows make real-image labels only approximate colors, we complement this supervision with pure-color anchors whose pixels exactly match their paired hex values. These anchors are used only at high-noise timesteps, leaving low-noise training to natural images. We further introduce Any Color Benchmark (ACBench), comprising ACBench-T2I and ACBench-Edit, to measure object-level hex color fidelity across both tasks. On FLUX.2-4B, Paint-Anything improves ACBench-T2I and ACBench-Edit scores by 85.3% and 28.3%, respectively, relative to the base model, with ablations supporting the training recipe. It also achieves the highest average CompColor score among the compared methods. --- ### Paint-Anything: Unified Any-Color Control for Image Generation and Editing Authors: Ji Xie, Dewei Zhou, Xinyu Huang, Zhennan Chen, Xun Wang Published: 2026-09-17 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.20816v1 Professional design requires any-color control: the ability to specify an object's target color with any 24-bit hex value for image generation and editing. Prior work has explored color generation, editing, and colorization, but often relies on dedicated color representations or specialized inference procedures. Advances in large language models offer a simpler starting point: even compact models can associate hex values with color semantics. We present Paint-Anything, which learns a shared hex-prompt interface for generation and editing through object-level color supervision. We develop a data pipeline that constructs Paint-500K from real images through object grounding, perceptual color labeling, and editing-pair synthesis. Since shadows make real-image labels only approximate colors, we complement this supervision with pure-color anchors whose pixels exactly match their paired hex values. These anchors are used only at high-noise timesteps, leaving low-noise training to natural images. We further introduce Any Color Benchmark (ACBench), comprising ACBench-T2I and ACBench-Edit, to measure object-level hex color fidelity across both tasks. On FLUX.2-4B, Paint-Anything improves ACBench-T2I and ACBench-Edit scores by 85.3% and 28.3%, respectively, relative to the base model, with ablations supporting the training recipe. It also achieves the highest average CompColor score among the compared methods. --- ### ERCPMP-Gx: Endoscopic Image and Video Dataset for Morphological, Histopathological, and Genomic Characterization of Colorectal Polyposis Authors: Zahra Ghaffari, Massih Bahar, Mojgan Forootan, Ali Darvishi, Hamidreza Bolhasani Published: 2026-09-17 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.20815v1 Hereditary polyposis syndromes can be precursor lesions to colorectal cancer and are associated with a broad spectrum of extracolonic tumors. Early identification and accurate classification of these syndromes are essential for timely diagnosis, individualized patient management, and targeted surveillance strategies for affected families. However, public endoscopic datasets are largely organized around the individual sporadic polyp, and none links the polyposis phenotype to histopathology and germline findings at the patient level. Here, we present ERCPMP-Gx, an endoscopic, histopathological, and genomic dataset developed to support the application of artificial intelligence (AI) in the recognition, characterization, and classification of colorectal polyposis. Most procedures were performed using the Olympus EVIS X1 system with white-light endoscopy (WLE), narrow-band imaging (NBI), magnifying NBI (M-NBI), and NBI with near focus modes, yielding 160 images and accompanying video clips. Approximately eighty percent of cases represent clinically and/or genetically confirmed hereditary polyposis syndromes (PG), including familial adenomatous polyposis (FAP), Peutz-Jeghers syndrome (PJS), juvenile polyposis syndrome (JPS), and ganglioneuroma syndrome (GNS), while the remaining twenty percent comprise non-hereditary polyps and polyp-mimicking lesions with overlapping morphological features (Non-PG), included to support differential classification. Each released record is linked, where available, to standardized endoscopic annotations, representative histopathology, and clinically reported germline findings, forming an AI-ready, patient-level annotation framework. The dataset is publicly accessible at Mendeley (https://doi.org/10.17632/nzyfc544bx.2). For the latest updates and further information, readers are referred to the DataBioX website: https://databiox.com. --- ### How Does Distribution Shift Shape Pretraining Gains in Neural PDE Surrogates? Authors: Pochinapeddi Sai Bhargav, Nithin Somasekharan, Rohit Sunil Kanchi, Sicheng He, Shaowu Pan Published: 2026-09-17 Categories: physics.comp-ph, cs.LG, physics.flu-dyn arXiv: https://arxiv.org/abs/2609.20814v1 Pretraining a neural PDE surrogate can reduce the amount of new CFD data needed when geometry or modeled physics changes. However, it remains unclear how different components of distribution shift affect this benefit. We pretrain a surrogate on 254,909 RANS solutions from one airfoil family and fine-tune it on a new family under two target settings with matched freestream ranges: the same Spalart-Allmaras (SA) modeling and SA with added $e^N$ transition modeling. At $N=1000$, the pretrained model matches the accuracy of a model trained from scratch on $3.25\times$ as many samples for the same-SA target, but $2.58\times$ as many for the transition-modeled target. By $N=5000$, this ordering reverses ($1.56\times$ versus $1.86\times$). At $N=1000$, sampling more distinct airfoils lowers error on both targets, but only for the same-SA target is the gain increase larger than the observed draw-to-draw variation ($3.3\times$ to $4.0\times$). These results show that pretraining value depends jointly on target-data budget, target-data coverage, and whether source and target differ in modeled physics. --- ### Quantifying Overclaiming Propensity in Frontier LLM Agents Authors: Nolan Smyth, Yorguin-Jose Mantilla-Ramos, Pascal Jr Tikeng Notsawo, Saskia Helbling, Alberto Tosato, Mohamed Amine Merzouk, Nouha Dziri, Gauthier Gidel, Tommaso Tosato Published: 2026-09-17 Categories: cs.SE, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.20812v3 Frontier coding agents are increasingly trusted to work autonomously for long periods of time, yet what they actually did is often hard to tell from their final response. We quantify the propensity of such agents to overclaim task completion, which may mislead the user. We operationalize overclaiming as a final response that reports work that the agent's own transcript shows it did not do, for example, claiming to have read a file it never opened. This criterion requires no inference about intent and does not depend on whether the delivered work is correct; it asks only whether the reported work was done. We introduce OverclaimBench, an evaluation suite of five file-review scenarios with transcript-based coverage measurements and registered planted defects. We evaluate eight proprietary frontier models in their own production command-line interfaces and four open-weight models under a single fixed harness, and find that 1) agents fail to read every file they were asked to review in 67.9% of runs; 2) among these incomplete runs, agents are misleading 80.4% of the time (59-96% per model), either falsely claiming a complete review or leaving the gap undisclosed; 3) requiring delegation to subagents increases coverage, but a large majority of reviews that remain incomplete are still misleading; and 4) agents that falsely claim a complete review miss planted defects at about 1.8 times the rate of agents that read every file, showing that claims of completion can conceal substantive failures. Together, these results show that agents' final responses are not reliable accounts of their actions. --- ### Quantifying Overclaiming Propensity in Frontier LLM Agents Authors: Nolan Smyth, Yorguin-Jose Mantilla-Ramos, Pascal Jr Tikeng Notsawo, Saskia Helbling, Alberto Tosato, Mohamed Amine Merzouk, Nouha Dziri, Gauthier Gidel, Tommaso Tosato Published: 2026-09-17 Categories: cs.SE, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.20812v1 Frontier coding agents are increasingly trusted to work autonomously for long periods, yet an agent's final response is often the only account of that work a user sees. We quantify the propensity of frontier agents to \emph{overclaim} task completion, a misrepresentation that can mislead the user. An agent overclaims when its final response contradicts information in its context. This definition requires no inference about intent and is independent of task success. We introduce \emph{OverclaimBench}, an evaluation suite composed of five file-review scenarios, transcript-based coverage measurements, and registered planted defects. We evaluate eight proprietary frontier models in their own production command-line interfaces, and four open-weight models under a single fixed harness on OverclaimBench and find that 1) agents do not read all the files they were asked to review in 67.9\% of runs; 2) among runs where not all files are read, agents are \emph{misleading} 80.4\% of the time (59--96\% per model), either falsely claiming to have read all files or omitting that coverage is incomplete; 3) requiring delegation to subagents increased reading coverage, but among reviews that remained incomplete, a large majority were still misleading; and 4) agents that falsely claimed a complete review missed planted defects at about 1.8 times the rate of agents that read every file, showing that claims of completion can conceal substantive failures. Together, these results show that agents' final responses are not reliable accounts of their actions. --- ### Unifying Models of Intergroup Hostility in Online Discourse Authors: Patrick Gerard, Julia Mendelsohn, Kristina Lerman Published: 2026-09-17 Categories: cs.CL, cs.SI arXiv: https://arxiv.org/abs/2609.20808v1 Hostile rhetoric toward social groups can normalize exclusion and justify mistreatment, as well as contribute to rising polarization and political violence. Efforts to moderate hostile rhetoric in online speech draw on foundational theories in social and moral psychology, and political science. However, these theories were developed largely in parallel, often propose different and sometimes conflicting accounts of how hostility develops, and have rarely been tested against each other in real discourse. The result is a fragmented understanding of the rhetorical mechanisms of hostility, without a clear sense of how they appear, and relate to each other, in real-world discourse. Using 2.86 million posts from TikTok, Truth Social, and Twitter/X during the 2024 U.S. presidential election, we model the mechanisms of six foundational theories of intergroup hostility -- boundary construction, threat construction, scapegoating, negative evaluation, dehumanization, and action orientation -- within a common empirical framework to recover the broader organization of intergroup hostility rhetoric. Structurally, we find that boundary construction and threat construction anchor the system; temporally, we find that these mechanisms tend to follow a regular ordering: boundary construction, derogation, and action orientation tend to appear early; dehumanization and threat construction later; scapegoating latest. Mapping how these theoretical frameworks actually manifest in discourse bridges longstanding divisions across social science traditions and presents computational social science with a clearer empirical foundation for modeling intergroup hostility rhetoric beyond single-label detection. ---