# Research Papers — 2026-05 · bundle page 255 of 326 > Full-text markdown of papers #12701–12750 of 16266 published in 2026-05, newest first. AI-readable mirrors of the full 16266-paper 2026-05 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-05/page/255 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/254.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/256.md) ### Convergent Stochastic Training of Multi-Headed Attention and Understanding LoRA Authors: Zhengkai Sun, Dibyakanti Kumar, Alejandro F Frangi, Anirbit Mukherjee, Mingfei Sun Published: 2026-05-08 Categories: cs.LG, math.FA, math.PR arXiv: https://arxiv.org/abs/2605.07959v2 Transformers have revolutionized machine learning and deploying attention layers in the model is increasingly standard across a myriad of applications. Further, for large models, it is common to implement Low Rank Adaptation (LoRA), whereby a factorized parameterization of them is trained, to achieve a surprisingly beneficial accuracy-size trade-off. In this work, via a unified framework we rigorously establish trainability of such models under stochastic methods. We prove that for a class of mild regularizations, the empirical regression loss on a attention layer and LoRA on a shallow neural net, both induce Poincaré inequality for the corresponding Gibbs' measure. Crucially, we show that the Poincaré constant is free of the data dimension for LoRA and for multi-head attention it is free of the head dimensions, Then it follows, via invoking recent results, that a certain SDE, which mimics the SGD, minimizes the corresponding losses. In both the cases, our first-of-its-kind results of trainability on attention and nets, do not use any assumptions on the data or the model size. --- ### Convergent Stochastic Training of Attention and Understanding LoRA Authors: Zhengkai Sun, Dibyakanti Kumar, Alejandro F Frangi, Anirbit Mukherjee, Mingfei Sun Published: 2026-05-08 Categories: cs.LG, math.FA, math.PR arXiv: https://arxiv.org/abs/2605.07959v1 Transformers have revolutionized machine learning and deploying attention layers in the model is increasingly standard across a myriad of applications. Further, for large models, it is common to implement Low Rank Adaptation (LoRA), whereby a factorized parameterization of them is trained, to achieve a surprisingly beneficial accuracy-size trade-off. In this work, via a unified framework we rigorously establish trainability of such models under stochastic methods. We prove that for any mild regularization, the empirical regression loss on a attention layer and LoRA on a shallow neural net, both induce Poincaré inequality for the corresponding Gibbs' measure. Then it follows via invoking recent results that a certain SDE, which mimics the SGD, minimizes the corresponding losses. In both the cases, our first-of-its-kind results of trainability on attention and nets, do not rely on any assumptions on the data or the size of the architecture. --- ### TimeLesSeg: Unified Contrast-Agnostic Cross-Sectional and Longitudinal MS Lesion Segmentation via a Stochastic Generative Model Authors: Vicent Caselles-Ballester, Eloy Martínez-Heras, Giuseppe Pontillo, Zoe Mendelsohn, Elena M. Marrón, Juan Luis García Fernández, Laia Subirats, Jon Stutters, Jeremy Chataway, Frederik Barkhof, Sara Llufriu, Ferran Prados Published: 2026-05-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.07955v1 Multiple sclerosis (MS) expresses substantial clinical and radiological heterogeneity, which poses significant challenges for automatic lesion segmentation. The current deep learning-based SOTA is highly susceptible to changes in both distribution, e.g., changes in scanner; as well as the structure of inputs, evident in the current divide between cross-sectional and longitudinal approaches. We introduce TimeLesSeg, a unified contrast-agnostic framework designed to segment MS lesions regardless of the presence of a temporal dimension in its inputs, with a single convolutional neural network. Our approach models pathological priors through lesion masks, which are processed together with the current scan. Cross-sectional processing is enabled by exposing the model to training cases where no prior information is available, which are modeled with an empty mask, allowing it to operate seamlessly in both scenarios. To overcome the scarcity and inconsistency of longitudinal datasets, we propose a novel generative pipeline in which patterns of lesion evolution are simulated by stochastically deforming each individual lesion with morphological operations, producing realistic prior timepoints. In parallel, we achieve contrast agnosticism through Gaussian mixture model-based domain randomization, enabling the network to experience a wide spectrum of intensity profiles. Results on three publicly available and two in-house datasets show that TimeLesSeg outperforms the contrast-agnostic state of the art on single-modality inputs across overlap- and distance-based metrics. In longitudinal processing, our method outperforms SAMSEG, and captures lesion load dynamics more accurately than both the former and LST-AI. All source code related to the development of TimeLesSeg is available at https://github.com/NeuroADaS-Lab/TimeLesSeg. --- ### Slowly Annealed Langevin Dynamics: Theory and Applications to Training-Free Guided Generation Authors: Atsushi Nitanda, Dake Bu, Yueming Lyu, Tanya Veeravalli Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07950v1 We study Slowly Annealed Langevin Dynamics (SALD), a sampler for tracking a path of moving target distributions and approximating the terminal target through time slowdown. We establish non-asymptotic convergence guarantees via a KL differential inequality, showing that slowdown improves tracking through contraction of intermediate targets and the complexity of the path. Motivated by training-free guided generation with pretrained score-based generative models, we further introduce Velocity-Aware SALD (VA-SALD), which explicitly incorporates the underlying marginal distributions of the pretrained model and uses slowdown to correct the additional deviation induced by guidance. This yields a principled framework for training-free guided generation for diffusion-based and related generative model families, together with convergence guarantees that clarify the roles of intermediate functional inequalities and guidance bias. Code is available at https://github.com/anitan0925/sald. --- ### Exploring the non-convexity in machine learning using quantum-inspired optimization Authors: Kandula Eswara Sai Kumar, Parth Dhananjay Danve, Abhishek Chopra, Rut Lineswala Published: 2026-05-08 Categories: cs.CE, cs.AI, cs.LG, math.OC arXiv: https://arxiv.org/abs/2605.07947v1 The escalating complexity of modern machine learning necessitates solving challenging non-convex optimization problems, particularly in high-dimensional regimes and scenarios contaminated by gross outliers. Traditional approaches, relying on convex relaxations or specialized local search heuristics, frequently succumb to suboptimal local minima and fail to recover the true underlying discrete structures. In this paper, we propose treating these non-convex challenges as a global search problem and introduce a unified framework based on Quantum-Inspired Evolutionary Optimization (QIEO). By leveraging a probabilistic representation inspired by quantum superposition, QIEO maintains a global view of the search space, enabling it to tunnel through local optima that trap conventional gradient-based and greedy solvers. We comprehensively evaluate QIEO across diverse non-convex applications, including sparse signal recovery (gene expression analysis and compressed sensing) and robust linear regression. Extensive benchmarking against state-of-the-art continuous solvers (ADAM, Differential Evolution), classical metaheuristics (Genetic Algorithms), and specialized non-convex algorithms (Iterative Hard Thresholding) demonstrates that QIEO consistently achieves superior structural fidelity, lower mean squared error, and enhanced robustness without support inflation. Our findings suggest that embracing a quantum-inspired global search provides a resilient, unified paradigm for overcoming the inherent intractability of discrete nonconvex machine learning landscapes. --- ### Rebalancing gradient to improve self-supervised co-training of depth, odometry and optical flow predictions Authors: Marwane Hariat, Antoine Manzanera, David Filliat Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07945v1 We present CoopNet, an approach that improves the cooperation of co-trained networks by dynamically adapting the apportionment of gradient, to ensure equitable learning progress. It is applied to motion-aware self-supervised prediction of depth maps, by introducing a new hybrid loss, based on a distribution model of photo-metric reconstruction errors made by, on the one hand the depth + odometry paired networks, and on the other hand the optical flow network. This model essentially assumes that the pixels from moving objects (that must be discarded for training depth and odometry), correspond to those where the two reconstructions strongly disagree. We justify this model by theoretical considerations and experimental evidences. A comparative evaluation on KITTI and CityScapes datasets shows that CoopNet improves or is comparable to the state-of-the-art in depth, odometry and optical flow predictions. --- ### TAVIS: A Benchmark for Egocentric Active Vision and Anticipatory Gaze in Imitation Learning Authors: Giacomo Spigler Published: 2026-05-08 Categories: cs.RO, cs.AI, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2605.07943v1 Active vision -- where a policy controls its own gaze during manipulation -- has emerged as a key capability for imitation learning, with multiple independent systems demonstrating its benefits in the past year. Yet there is no shared benchmark to compare approaches or quantify what active vision contributes, on which task types, and under what conditions. We introduce TAVIS, evaluation infrastructure for active-vision imitation learning, with two complementary task suites -- TAVIS-Head (5 tasks, global search via pan/tilt necks) and TAVIS-Hands (3 tasks, local occlusion via wrist cameras) -- on two humanoid torso embodiments (GR1T2, Reachy2), built on IsaacLab. TAVIS provides three evaluation primitives: a paired headcam-vs-fixedcam protocol on identical demonstrations; GALT (Gaze-Action Lead Time), a novel metric grounded in cognitive science and HRI that quantifies anticipatory gaze in learned policies; and procedural ID/OOD splits. Baseline experiments with Diffusion Policy and $π_0$ reveal that (i) active-vision generally helps, but benefits are task-conditional rather than uniform; (ii) multi-task policies degrade sharply under controlled distribution shifts on both suites; and (iii) imitation alone yields anticipatory gaze, with median lead times comparable to the human teleoperator reference. Code, evaluation scripts, demonstrations (LeRobot v3.0; ~2200 episodes) and trained baselines are released at https://github.com/spiglerg/tavis and https://huggingface.co/tavis-benchmark. --- ### Delta-Adapter: Scalable Exemplar-Based Image Editing with Single-Pair Supervision Authors: Jiacheng Chen, Songze Li, Han Fu, Baoquan Zhao, Wei Liu, Yanyan Liang, Li Qing, Xudong Mao Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07940v1 Exemplar-based image editing applies a transformation defined by a source-target image pair to a new query image. Existing methods rely on a pair-of-pairs supervision paradigm, requiring two image pairs sharing the same edit semantics to learn the target transformation. This constraint makes training data difficult to curate at scale and limits generalization across diverse edit types. We propose Delta-Adapter, a method that learns transferable editing semantics under single-pair supervision, requiring no textual guidance. Rather than directly exposing the exemplar pair to the model, we leverage a pre-trained vision encoder to extract a semantic delta that encodes the visual transformation between the two images. This semantic delta is injected into a pre-trained image editing model via a Perceiver-based adapter. Since the target image is never directly visible to the model, it can serve as the prediction target, enabling single-pair supervision without requiring additional exemplar pairs. This formulation allows us to leverage existing large-scale editing datasets for training. To further promote faithful transformation transfer, we introduce a semantic delta consistency loss that aligns the semantic change of the generated output with the ground-truth semantic delta extracted from the exemplar pair. Extensive experiments demonstrate that Delta-Adapter consistently improves both editing accuracy and content consistency over four strong baselines on seen editing tasks, while also generalizing more effectively to unseen editing tasks. Code will be available at https://delta-adapter.github.io. --- ### Prototype Guided Post-pretraining for Single-Cell Representation Learning Authors: Sachini Weerasekara, Natasha Darras, Sagar Kamarthi, Colles Price, Jacqueline Isaacs Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07938v1 Single-cell representation learning (SCRL) from gene expression data offers a way to uncover the complex regulatory logic underlying cellular function. Inspired by large language models in natural language modeling, several single-cell pretrained models have recently been proposed that treat genes as tokens and cells as sentences. However, these models are fundamentally limited by the long-tailed nature of cell-type distributions and struggle to generalize under covariate shifts in gene expression data. While fine-tuning is often used to mitigate these issues, we observe that performance remains bounded. To address this challenge, we introduce CellRefine, a post-pretraining method that operates between the pretraining and fine-tuning stages of a single-cell foundation model. CellRefine uses a multi-faceted objective that incorporates marker-gene sets as structural priors to guide post-pretraining and refine the latent embedding manifold of cells. Across multiple computational biology tasks, empirical results show that CellRefine consistently improves downstream performance, yielding gains up to 15%. --- ### Ask Early, Ask Late, Ask Right: When Does Clarification Timing Matter for Long-Horizon Agents? Authors: Anmol Gulati, Hariom Gupta, Elias Lumer, Sahil Sen, Vamse Kumar Subbiah Published: 2026-05-08 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.07937v1 Long-horizon AI agents execute complex workflows spanning hundreds of sequential actions, yet a single wrong assumption early on can cascade into irreversible errors. When instructions are incomplete, the agent must decide not only whether to ask for clarification but when, and no prior work measures how clarification value changes over the course of execution. We introduce a forced-injection framework that provides ground-truth clarifications at controlled points in the agent's trajectory across four information dimensions (goal, input, constraint, context), three agent benchmarks, and four frontier models (three per benchmark; one on a single benchmark only; 84 task variants; 6,000+ runs). Counter to the common intuition that "earlier is always better," we find that the value of clarification depends sharply on what information is missing: goal clarification loses nearly all value after 10% of execution (pass@3 drops from 0.78 to baseline), while input clarification retains value through roughly 50%. Deferring any clarification type past mid-trajectory degrades performance below never asking at all. Cross-model Kendall tau correlations (0.78-0.87 among models sharing identical task coverage; 0.34-0.67 across the full 4-model panel) confirm these timing profiles are substantially task-intrinsic. A complementary study of 300 unscripted sessions reveals that no current frontier model asks within the empirically optimal window, with strategies ranging from over-asking (52% of sessions) to never asking at all. These empirical demand curves provide the quantitative foundation that existing theoretical frameworks require but have lacked, and establish concrete design targets for timing-aware clarification policies. Code and data will be publicly released. --- ### TraceFix: Repairing Agent Coordination Protocols with TLA+ Counterexamples Authors: Shuren Xia, Qiwei Li, Taqiya Ehsan, Jorge Ortiz Published: 2026-05-08 Categories: cs.AI, cs.MA arXiv: https://arxiv.org/abs/2605.07935v1 We present TraceFix, a verification-first pipeline for Large Language Model (LLM) multi-agent coordination. An agent synthesizes a protocol topology as a structured intermediate representation (IR) from a task description, generates PlusCal coordination logic, and iteratively repairs the protocol using counterexamples from the TLA+ model checker (TLC) until verification succeeds. Verified process bodies are compiled into per-agent system prompts and executed under a runtime monitor that rejects out-of-topology coordination operations. On 48 tasks spanning 16 scenario families, all tasks reach full TLC verification; 62.5% pass on the first attempt and none requires more than four repair iterations. State spaces span six orders of magnitude yet verification completes in under 60 s for every task. A 3,456-run runtime comparison shows that topology-monitored execution achieves the highest task completion (89.4% average, 81.5% full) and that runtimes using the verified protocol degrade at roughly half the rate of prompt-only and chat-only baselines when model capability is reduced. A paired ablation under a fixed runtime shows that TLC-verified protocols cut deadlock/livelock (DL/LL) from 31.1% to 14.1%, with the largest separation under fault injection. --- ### How to Train Your Latent Diffusion Language Model Jointly With the Latent Space Authors: Viacheslav Meshchaninov, Alexander Shabalin, Egor Chimbulatov, Nikita Gushchin, Ilya Koziev, Alexander Korotin, Dmitry Vetrov Published: 2026-05-08 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.07933v1 Latent diffusion models offer an attractive alternative to discrete diffusion for non-autoregressive text generation by operating on continuous text representations and denoising entire sequences in parallel. The major challenge in latent diffusion modeling is constructing a suitable latent space. In this work, we present the Latent Diffusion Language Model (LDLM), in which the latent encoder, diffusion model, and decoder are trained jointly. LDLM builds its latent space by reshaping the representations of a pre-trained language model with a trainable encoder, yielding latents that are easy to both denoise and decode into tokens. We show that naive joint training produces a low-quality diffusion model, and propose a simple training recipe consisting of an MSE decoder loss, diffusion-to-encoder warmup, adaptive timestep sampling, and decoder-input noise. Ablations show that each component substantially impacts generation performance. On OpenWebText and LM1B, LDLM achieves better generation performance than existing discrete and continuous diffusion language models while being $2{\text -}13\times$ faster, indicating that jointly learning the latent space is a key step toward making latent diffusion competitive for text generation. --- ### One Token Per Frame: Reconsidering Visual Bandwidth in World Models for VLA Policy Authors: Zuojin Tang, Shengchao Yuan, Xiaoxin Bai, Zhiyuan Jing, De Ma, Gang Pan, Bin Liu Published: 2026-05-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.07931v4 Vision-language-action (VLA) models can use visual prediction to anticipate future states, but dense visual features make the generative sequence grow with the number of camera views, prediction horizon, and encoder resolution. Whether such dense representations are necessary for effective control remains unclear. We introduce OneWM-VLA, which represents each retained camera view with one predictive token per future step. Adaptive Attention Pooling compresses visual features into compact latents, which are jointly generated with robot actions under a conditional flow-matching objective. Future observations provide the latent targets during training and are not required at inference. This design incorporates visual prediction into a pretrained VLA policy while keeping the generative sequence compact. On MetaWorld~MT50, OneWM-VLA improves the average success rate of the $π_0$ backbone from $47.91\%$ to $61.53\%$, reaching $72.01\%$ after 60k training steps. It also achieves $98.1\%$ success on LIBERO and raises Fold Cloth success on a real Piper arm from $20.0\%$ to $60.0\%$ relative to $π_0$. Comparisons on two additional VLA backbones consistently favor one token over three across the evaluated checkpoints. A matched ablation at a longer action horizon further shows that removing the latent loss reduces success from $58.09\%$ to $21.64\%$, supporting the benefit of future supervision for policy learning. --- ### One Token Per Frame: Reconsidering Visual Bandwidth in World Models for VLA Policy Authors: Zuojin Tang, Shengchao Yuan, Xiaoxin Bai, Zhiyuan Jing, De Ma, Gang Pan, Bin Liu Published: 2026-05-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.07931v3 Vision-language-action (VLA) models increasingly rely on auxiliary world modules to plan over long horizons, yet how such modules should be parameterized on top of a pretrained VLA remains an open design question. Existing world-model-augmented VLAs typically pass the per-frame visual stream into the world module at high visual bandwidth and treat its rollout as a side product of action prediction; under a constrained adaptation budget on a frozen backbone, this leaves both the per-frame representation and the latent action coupling under-examined. We introduce OneWM-VLA, which compresses each view into a single semantic token per frame through an Adaptive Attention Pooling, and produces the resulting latent stream and the action trajectory under a single flow-matching objective rather than connecting them through a separate decoder. Empirically, we find that per-frame visual bandwidth can be reduced to a single token without compromising long-horizon performance under our setup. Trained with 14.71M LoRA parameters on a $π_0$ (2B) backbone, OneWM-VLA improves the average success rate from 47.9% to 61.3% on MetaWorld~MT50, reaches 95.6% on LIBERO-Long (vs.85.2% for $π_0$), and reaches 60.0% on the long-horizon deformable task Fold Cloth on a real Piper arm (vs.20.0% for $π_0$). --- ### INO-SGD: Addressing Utility Imbalance under Individualized Differential Privacy Authors: Xiao Tian, Jue Fan, Rachael Hwee Ling Sim, Bryan Kian Hsiang Low Published: 2026-05-08 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.07930v1 Differential privacy (DP) is widely employed in machine learning to protect confidential or sensitive training data from being revealed. As data owners gain greater control over their data due to personal data ownership, they are more likely to set their own privacy requirements, necessitating individualized DP (IDP) to fulfil such requests. In particular, owners of data from more sensitive subsets, such as positive cases of stigmatized diseases, likely set stronger privacy requirements, as leakage of such data could incur more serious societal impact. However, existing IDP algorithms induce a critical utility imbalance problem: Data from owners with stronger privacy requirements may be severely underrepresented in the trained model, resulting in poorer performance on similar data from subsequent users during deployment. In this paper, we analyze this problem and propose the INO-SGD algorithm, which strategically down-weights data within each batch to improve performance on the more private data across all iterations. Notably, our algorithm is specially designed to satisfy IDP, while existing techniques addressing utility imbalance neither satisfy IDP nor can be easily adapted to do so. Lastly, we demonstrate the empirical feasibility of our approach. --- ### AgentEscapeBench: Evaluating Out-of-Domain Tool-Grounded Reasoning in LLM Agents Authors: Zhengkang Guo, Yiyang Li, Lin Qiu, Xiaohua Wang, Jingwen Xv, Dongyu Ru, Xiaoyu Li, Xiaoqing Zheng, Xuezhi Cao, Xunliang Cai Published: 2026-05-08 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.07926v2 As LLM-based agents increasingly rely on external tools, it is important to evaluate their ability to sustain tool-grounded reasoning beyond familiar workflows and short-range interactions. We introduce AgentEscapeBench, an escape-room-style benchmark that tests whether agents can infer, execute, and revise novel tool-use procedures under explicit long-range dependency constraints. Each task defines a directed acyclic dependency graph over tools and items, requiring agents to invoke real external functions, track hidden state revealed incrementally, propagate intermediate results, and submit a deterministically verifiable final answer. AgentEscapeBench includes 270 instances across five difficulty tiers and supports fully automated evaluation. Experiments with sixteen LLM agents and human participants show that performance drops sharply as dependency depth increases: humans decline from 98.3% success at difficulty-5 to 80.0% at difficulty-25, while the best model drops from 90.0% to 60.0%. Trajectory analysis attributes model failures mainly to breakdowns in long-range state tracking, clue adherence, and intermediate-result propagation. These findings suggest that current agents can often handle local tool use but still struggle with deep contextual dependencies. We hope AgentEscapeBench can serve as a diagnostic testbed for measuring current agent capabilities and informing future training efforts toward more robust general-purpose reasoning, action, and adaptation. --- ### How Value Induction Reshapes LLM Behaviour Authors: Arnav Arora, Natalie Schluter, Katherine Metcalf, Maartje ter Hoeve Published: 2026-05-08 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.07925v1 Conversational Large Language Models are post-trained on language that expresses specific behavioural traits, such as curiosity, open-mindedness, and empathy, and values, such as helpfulness, harmlessness, and honesty. This is done to increase utility, ensure safety, and improve the experience of the people interacting with the model. However, values are complex and inter-related -- inducing one could modify behaviour on another. Further, inducing certain values can make models more addictive or sycophantic through language used in the generations, with a potential detrimental effect on the user. We investigate these and other unintended effects of value induction into models. We fine-tune models using curated value subsets of existing preference datasets, measuring the impact of value induction on expression of other values, model safety, anthropomorphic language, and various QA benchmarks. We find that (i) inducing values leads to expression of other related, and sometimes contrastive values, (ii) inducing positive values increases safety, and (iii) all values increase anthropomorphic language use, making models more validating and sycophantic. --- ### Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillation Authors: Amin Karimi Monsefi, Dominic Culver, Nikhil Bhendawade, Manuel R. Ciosici, Yizhe Zhang, Irina Belousova Published: 2026-05-08 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2605.07924v1 Discrete flow matching generates text by iteratively transforming noise tokens into coherent language, but may require hundreds of forward passes. Distillation uses the multi-step trajectory to train a student to reproduce the process in a few steps. When the student underperforms, the usual explanation is insufficient capacity. We argue the opposite: the trajectory is the bottleneck, not the student. Each training trajectory is built through a chain of blind stochastic jumps with no evaluation of sequence quality; a single bad decision at an early midpoint propagates through subsequent steps, yet the student must imitate the result. Trajectory-Shaped Discrete Flow Matching (TS-DFM) replaces these blind jumps with guided navigation: a lightweight energy compass evaluates candidate continuations at each midpoint, selecting the most coherent. All shaping is training-only; inference cost is unchanged. On 170M-parameter language modeling, the shaped student at 8 steps achieves 32% lower perplexity than the 1,024-step teacher while being 128x faster, with gains consistent across source distributions and three evaluators of increasing scale. TS-DFM achieves the best perplexity of any discrete-generation baseline we compare against, including methods trained on 6x more data or using 5x larger models. --- ### Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders Authors: Tue M. Cao, Hoang X. Nhat, Raed Alharbi, Phi Le Nguyen, My T. Thai Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07922v2 Learning hierarchical features in Sparse Autoencoders (SAEs) is essential for capturing the structured nature of real-world data and mitigating issues like feature absorption or splitting. Existing works attempt to identify hierarchical relationships within independent feature sets by relying on activation coverage, the assumption that child feature should only activate when its parent feature activates. However, we demonstrate that this condition alone is insufficient; that is, it often produces false positives where parent and child concepts are semantically unrelated. To address this, we introduce a novel reconstruction condition that enforces a deeper functional link between hierarchical levels. By combining both activation and reconstruction constraints, we propose the Tree SAE, a model designed to learn hierarchical structures directly from within the feature set. Our results demonstrate that Tree SAEs significantly surpass the existing SAEs at learning hierarchical pairs while maintaining competitive performance to the state-of-the-art on several key benchmarks. Finally, we demonstrate the practical utility of our Tree SAE in mapping the geometry of child feature subspaces and uncovering the complex hierarchical concept structures encoded within large language models. --- ### MedVIGIL: Evaluating Trustworthy Medical VLMs Under Broken Visual Evidence Authors: Hanqi Jiang, Junhao Chen, Mingyu Kang, Hyeokjae Kwon, Yi Pan, Lifeng Chen, Weihang You, Haozhen Gong, Ruiyu Yan, Jinglei Lv, Lin Zhao, Hui Ren et al. Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07919v2 Medical vision--language models (VLMs) are usually evaluated on intact image--question pairs, but trustworthy clinical use requires a stronger property: a model must recognise when the evidential basis for an answer has failed. We study this through silent failures under perturbed evidence, where a vision-required medical question is paired with a false premise, wording perturbation, knowledge-only rewrite, or ROI-corrupted image, yet the model returns a fluent non-refusal answer. We introduce medvigil, a 300-case evaluation suite drawn from four public medical VQA sources, supervised end to end by four board-certified radiologists: every gold answer, refusal option, candidate-answer set, paraphrase, false-premise trap, ROI box, and clinical risk tier is clinician-authored. Two attending radiologists annotate every case in parallel, a senior radiologist consolidates the released manifest, and a separate fourth radiologist independent of construction answers every probe to provide the human reference baseline. The release contains 2556 MCQ probes, 240 counterfactual triplets, physician-adjudicated risk-tier and answerability flags, ROI boxes, and a paired open-ended variant. We report seven correctness-conditioned audit metrics that summarise into the medvigil Composite Score (MCS), and audit 16 vision-capable models plus two text-only baselines. The independent radiologist scores MCS 83.3 at silent-failure rate 5.8%, leaving a 14.1-point composite headroom above the strongest audited model (Claude Opus 4.7 at 69.2). The benchmark and evaluation harness are publicly released. --- ### What Matters for Diffusion-Friendly Latent Manifold? Prior-Aligned Autoencoders for Latent Diffusion Authors: Zhengrong Yue, Taihang Hu, Mengting Chen, Haiyu Zhang, Zihao Pan, Tao Liu, Zikang Wang, Jinsong Lan, Xiaoyong Zhu, Bo Zheng, Yali Wang Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07915v1 Tokenizers are a crucial component of latent diffusion models, as they define the latent space in which diffusion models operate. However, existing tokenizers are primarily designed to improve reconstruction fidelity or inherit pretrained representations, leaving unclear what kind of latent space is truly friendly for generative modeling. In this paper, we study this question from the perspective of latent manifold organization. By constructing controlled tokenizer variants, we identify three key properties of a diffusion-friendly latent manifold: coherent spatial structure, local manifold continuity, and global manifold semantics. We find that these properties are more consistent with downstream generation quality than reconstruction fidelity. Motivated by this finding, we propose the Prior-Aligned AutoEncoder (PAE), which explicitly shapes the latent manifold instead of leaving diffusion-friendly manifold to emerge indirectly from reconstruction or inheritance. Specifically, PAE leverages refined priors derived from VFMs and perturbation-based regularization to turn spatial structure, local continuity, and global semantics into explicit training objectives. On ImageNet 256x256, PAE improves both training efficiency and generation quality over existing tokenizers, reaching performance comparable to RAE with up to 13x faster convergence under the same training setup and achieving a new state-of-the-art gFID of 1.03. These results highlight the importance of organizing the latent manifold for latent diffusion models. --- ### Flatness and Gradient Alignment Are Both Necessary: Spectral-Aware Gradient-Aligned Exploration for Multi-Distribution Learning Authors: Aristotelis Ballas, Christos Diou Published: 2026-05-08 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2605.07914v2 Sharpness-aware and gradient-alignment methods have been shown to improve generalization, however each family of methods targets a single geometric property of the loss landscape, while ignoring the other. In this paper, we show that this omission is structurally unavoidable and that both flatness and gradient alignment should be considered in multi-distribution learning settings. Specifically, we derive an excess-risk decomposition that yields two additive leading-order terms: (i) an alignment term, controlled by the trace of $\bar{H}^{-1}Σ_g$ and (ii) a curvature term, controlled by $\bar{H}$, where $\bar{H}$ is the average Hessian and $Σ_g$ is the covariance of the gradient across distributions. Notably, $\bar{H}$ appears inverted in one and non-inverted in the other. We further show, via a counterexample, that neither quantity bounds the other in general, so no algorithm targeting only one term can guarantee low excess risk. Motivated by this decomposition, we propose SAGE (Spectral-Aware Gradient-Aligned Exploration) that targets both terms. The curvature component replaces SAM's gradient-scaled perturbation with the polar factor of each layer's gradient matrix, computed via Newton-Schulz iteration, so that the ascent step probes all directions with similar magnitude. On the other hand, the alignment component injects isotropic noise at the descent step, the magnitude of which scales with cross-distribution gradient disagreement. Experiments on five domain-generalization and two multi-task learning benchmarks show that the proposed method establishes a new state-of-the-art on DomainBed and acts as a general-purpose improvement to base MTL solvers, remaining competitive with, or even surpassing, state-of-the-art methods. --- ### Sycophantic AI makes human interaction feel more effortful and less satisfying over time Authors: Lujain Ibrahim, Franziska Sofia Hafner, Myra Cheng, Cinoo Lee, Rebecca Anselmetti, Robb Willer, Luc Rocher, Diyi Yang Published: 2026-05-08 Categories: cs.HC, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2605.07912v3 Millions of people now turn to artificial intelligence (AI) systems for personal advice, guidance, and support. Such systems can be sycophantic, frequently affirming users' views and beliefs. Across five preregistered studies (N = 3,075 participants, 12,766 human-AI conversations), including a three-week study with a census-representative U.S. sample, we provide longitudinal experimental evidence that sycophantic AI shifts how users approach their closest relationships. We show that sycophantic AI immediately delivers the emotional and esteem support users typically associate with close friends and family. Over three weeks of such interactions, users became nearly as likely to seek personal advice from sycophantic AI as from close friends and family, and reported lower satisfaction with their real-world social interactions. When given a choice among AI response styles, a majority preferred sycophantic AI -- not for the quality of its advice, but because it made them feel most understood. Together, these findings offer a relational account of AI sycophancy and its impacts. --- ### One World, Dual Timeline: Decoupled Spatio-Temporal Gaussian Scene Graph for 4D Cooperative Driving Reconstruction Authors: Yulong Chen, Xiaoyun Dong, Haoyu Zhang, Zongxian Yang, Lewei Xie, Xinke Li, Yifan Zhang, Kai Wang, Jianping Wang Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07910v2 Reconstructing dynamic scenes from Vehicle-to-Infrastructure Cooperative Autonomous Driving (VICAD) data is fundamentally complicated by temporal asynchrony: vehicle and infrastructure cameras operate on independent clocks, capturing the same dynamic agent such as cars and pedestrians at different physical times. Existing Gaussian Scene Graph methods implicitly assume synchronized observations and assign a single pose per agent per frame, which is an assumption that breaks in cooperative settings, where the resulting gradient conflicts cause severe ghosting on dynamic agents. We identify this as a representation-level failure, not an optimization artifact: we prove that any single-timeline formulation incurs an irreducible photometric loss scaling quadratically with agent velocity and cross-source time offset. To resolve this, we propose Dust (DecoUpled Spatio-Temporal) Gaussian Scene Graph for 4D Cooperative Driving Reconstruction. DUST Gaussian Scene Graph shares a canonical Gaussian set per agent for appearance consistency, while maintaining decouple pose trajectories aligned to each source's true capture timestamps. We prove that this decoupling enables the pose-gradient kernel block-diagonal, eliminating cross-source interference entirely. To make Dust practical, we further introduce a static anchor-based pose correction pipeline that corrects spatio misalignment between vehicle and infrastructure annotations, and a pose-regularized joint optimization scheme that prevents trajectory jitter and drift during early training. On 26 sequences from V2X-Seq, DUST achieves state-of-the-art performance, improving dynamic-area PSNR by 3.2 dB over the strongest baseline and reducing Fréchet Video Distance by 37.7%, with keeping robustness under larger temporal asynchrony. --- ### Statistical inference with belief functions: A survey Authors: Fabio Cuzzolin Published: 2026-05-08 Categories: math.ST, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.07908v1 Belief functions are a powerful and popular framework for the mathematical characterisation of uncertainty, in particular in situations in which lack of data renders learning a probability distribution for the problem impractical. The first step in a reasoning chain based on belief functions is inference: how to learn a belief measure from the available data. In this survey we focus, in particular, on making inference from statistical data, and review the most significant contributions in the area. --- ### Consistency Regularised Gradient Flows for Inverse Problems Authors: Alessio Spagnoletti, Tim Y. J. Wang, Marcelo Pereyra, O. Deniz Akyildiz Published: 2026-05-08 Categories: stat.ML, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2605.07907v1 Vision-Language Latent Diffusion Models (LDMs) (Rombach et al., 2022) provide powerful generative priors for inverse problems. However, existing LDM-based inverse solvers typically require a large number of neural function evaluations (NFEs) and backpropagation through large pretrained components, leading to substantial computational costs and, in some cases, degraded reconstruction quality. We propose a unified Euclidean-Wasserstein-2 gradient-flow framework that jointly performs posterior sampling and prompt optimization in the latent space through a single flow that aligns the prior and posterior with the observed data. Combined with few-step latent text-to-image models, this formulation enables low-NFE inference without backpropagation through autoencoders. Experiments across several canonical imaging inverse problems show that our method achieves state-of-the-art performance with significantly reduced computational cost. --- ### CoCoReviewBench: A Completeness- and Correctness-Oriented Benchmark for AI Reviewers Authors: Hexuan Deng, Xiaopeng Ke, Yichen Li, Ruina Hu, Dehao Huang, Derek F. Wong, Yue Wang, Xuebo Liu, Min Zhang Published: 2026-05-08 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.07905v2 Despite the rapid development of AI reviewers, evaluating such systems remains challenging: metrics favor overlap with human reviews over correctness. However, since human reviews often cover only a subset of salient issues and sometimes contain mistakes, they are unreliable as gold references. To address this, we build category-specific benchmark subsets and skip evaluation when the corresponding human reviews are missing to strengthen Completeness. We also leverage reviewer--author--meta-review discussions as expert annotations and filter unreliable reviews accordingly to strengthen Correctness. Finally, we introduce CoCoReviewBench, which curates 3,900 papers from ICLR and NeurIPS to enable reliable and fine-grained evaluation of AI reviewers. Analysis shows that AI reviewers remain limited in correctness and are prone to hallucinations, and highlights reasoning models as more effective reviewers, motivating further directions for improving AI reviewers. Benchmarks and models are available at https://github.com/hexuandeng/CoCoReviewBench. --- ### BeeVe: Unsupervised Acoustic State Discovery in Honey Bee Buzzing Authors: Hamze Hammami, Nidhal Abdulaziz Published: 2026-05-08 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2605.07903v1 Discovering structure in biological signals without supervision is a fundamental problem in computational intelligence, yet existing bioacoustic methods assume vocal production models or predefined semantic units, leaving non-vocal species poorly served. This work introduces BeeVe, an unsupervised framework for acoustic state discovery in collective honey bee buzzing. BeeVe uses the self-supervised Patchout Spectrogram Transformer (PaSST) as a frozen feature extractor, then trains a Vector-Quantized Variational Autoencoder (VQ-VAE) without labels on those embeddings, learning a finite discrete codebook of acoustic tokens directly from unlabelled hive audio. No labels, pretext tasks, or contrastive objectives are used at any stage. Post-hoc evaluation against known queen status reveals that the learned tokens separate queenright and queenless conditions with Jensen-Shannon Divergence values between 0.609 and 0.688, and that the queenless condition further decomposes into three internally coherent sub-states stable across experiments with different codebook sizes and random seeds. Token transition analysis confirms non-random sequential structure (p << 0.001) across all experiments. Generalisation to unseen recordings preserves both token overlap (Jaccard = 0.947) and global manifold topology. These results demonstrate that unsupervised discrete codebook learning can recover repeatable acoustic structure from a non-vocal biological signal without annotation, opening a path toward non-invasive acoustic hive health monitoring. --- ### Curvature Beyond Positivity: Greedy Guarantees for Arbitrary Submodular Functions Authors: Yixin Chen, Alan Kuhnle Published: 2026-05-08 Categories: cs.LG, cs.DS arXiv: https://arxiv.org/abs/2605.07902v1 Submodular functions -- functions exhibiting diminishing returns -- are central to machine learning. When the objective is monotone and non-negative, the greedy algorithm achieves a tight $63\%$ approximation. But many practical objectives incorporate costs that make them negative on some inputs, and all existing multiplicative guarantees require non-negativity. Prior work handles negativity through additive bounds for the special class of decomposable functions and non-monotonicity through partial-monotonicity parameters, but these address each difficulty in isolation and neither extends the classical structural theory. We extend \emph{curvature} -- a parameter measuring how far a function deviates from linearity -- to all submodular functions, handling both non-monotonicity and negativity through a single classical concept. A greedy algorithm with pruning achieves a curvature-controlled multiplicative ratio for \emph{any} submodular function, including those taking negative values -- the first such guarantee beyond monotonicity and non-negativity. In the non-monotone regime $1 \le c_g < 2.2$, the bound strictly beats the best known uniform ratio of $0.401$ (for non-negative $f$), and it recovers the classical $(1-e^{-c_g})/c_g$ guarantee for monotone functions. A multilinear-extension variant extends the framework to general combinatorial constraints via multilinear relaxation. Experiments on cost-penalized experimental design, coverage, feature selection, and a curvature sweep on Multi-News passage selection support the theory. --- ### Semantic-Aware Adaptive Visual Memory for Streaming Video Understanding Authors: Hang Wu, Sherin Mary Mathews, Yujun Cai, Ming-Hsuan Yang, Yiwei Wang Published: 2026-05-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.07897v1 Online streaming video understanding requires models to process continuous visual inputs and respond to user queries in real time, where the unbounded stream and unpredictable query timing turn memory management into a central challenge. Existing methods typically compress visual tokens via visual similarity heuristics, or augment compression with KV-cache-level retrieval. However, compression decisions rarely incorporate semantic signals, and retrieval is often added after compression is finalized, making the two stages hard to coordinate. We present SAVEMem, a training-free dual-stage framework that brings semantic awareness into memory generation and lets the retrieval scope adapt per query. In Stage~1, SAVEMem builds a three-tier streaming memory online under a constant memory budget. A fixed pseudo-question bank provides a lightweight semantic prior, so that long-term retention is shaped by semantic salience rather than visual similarity alone. In Stage~2, SAVEMem performs query-aware retrieval over this memory. An anchor-conditioned recency gate adapts the retrieval scope from short-term to mid- and long-term memory based on whether the query targets the present or the distant past. Within this scope, late interaction between query and memory tokens selects candidate frames for answering. Applied to Qwen2.5-VL without training, SAVEMem improves the OVO-Bench overall score from 52.27 to 62.69 and yields consistent gains on StreamingBench and ODV-Bench, while reducing peak GPU memory by 48\% at 128 frames over the backbone. --- ### What if AI systems weren't chatbots? Authors: Sourojit Ghosh, Pranav Narayanan Venkit, Sanjana Gautam, Avijit Ghosh Published: 2026-05-08 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2605.07896v1 The rapid convergence of artificial intelligence (AI) toward conversational chatbot interfaces marks a critical moment for the industry. This paper argues that the chatbot paradigm is not a neutral interface choice, but a dominant sociotechnical configuration whose widespread adoption reshapes social, economic, legal, and environmental systems. We examine how treating AI primarily as conversational assistants has extensive structural downsides. We show how chatbot-based systems often fail to adequately meet user needs, particularly in complex or high-stakes contexts, while projecting confidence and authority. We further analyze how the normalization of chatbot-mediated interaction alters patterns of work, learning, and decision-making, contributing to deskilling, homogenization of knowledge, and shifting expectations of expertise. Finally, we examine broader societal effects, including labor displacement, concentration of economic power, and increased environmental costs driven by sustained investment in large-scale chatbot infrastructures. While acknowledging legitimate benefits, we argue that the current trajectory of AI development reflects specific value choices that prioritize conversational generality over domain specificity, accountability, and long-term social sustainability. We conclude by outlining alternative directions for AI development and governance that move beyond one-size-fits-all chatbots, emphasizing pluralistic system design, task-specific tools, and institutional safeguards to mitigate social and economic harm. --- ### Adaptive Regularization for Sparsity Control in Bregman-Based Optimizers Authors: Ahmad Aloradi, Tim Roith, Emanuël A. P. Habets, Daniel Tenbrinck Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07892v3 Sparse training reduces the memory and computational costs of deep neural networks. However, sparse optimization methods, e.g., those adding an $\ell_1$ penalty, often control sparsity only indirectly through a regularization parameter $λ$, whose mapping to the final sparsity rate is non-trivial. In our experiments, we found this parameter sensitivity to be particularly pronounced for Bregman-based optimizers. Specifically, the two variants LinBreg and AdaBreg reach the same sparsity at $λ$ values that differ by up to two orders of magnitude, requiring expensive trial-and-error sweeps to achieve a user-specified sparsity. To address this, we propose an adaptive regularization scheme that updates $λ$ based on the difference between the model's current sparsity and the target sparsity. We analyze the resulting algorithm and evaluate it on automatic speaker verification with ECAPA-TDNN and ResNet34 on VoxCeleb and CNCeleb. The proposed method reliably achieves sparsity targets ranging between 75% and 99%. It also converges faster than the oracle-tuned non-adaptive baseline during early training and matches or surpasses its final performance in equal error rate. We further show that the adaptive scheme inherits key properties from its non-adaptive counterpart, including improved out-of-distribution robustness over the dense baselines. --- ### Enhancing Federated Quadruplet Learning: Stochastic Client Selection and Embedding Stability Analysis Authors: Ozgu Goksu, Nicolas Pugeault Published: 2026-05-08 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2605.07888v1 Federated Learning (FL) enables decentralised model training across distributed clients without requiring data centralisation. However, the generalisation performance of the global model is usually degraded by data heterogeneity across clients, particularly under limited data availability and class imbalance. To address this challenge, we propose FedQuad, a novel method that explicitly enforces minimising intra-class representations while enabling inter-class splits across clients. By jointly minimising distances between positive pairs and maximising distances between negative pairs, the proposed approach mitigates representation misalignment introduced during model aggregation. We evaluate our method on CIFAR-10, CIFAR-100, and Tiny-ImageNet under diverse non-IID settings and varying numbers of clients, demonstrating consistent improvements over existing baselines. Additionally, we provide a comprehensive analysis of metric learning-based approaches in both centralised and federated environments, highlighting their effectiveness in alleviating representation collapse under heterogeneous data distributions. --- ### Characterizing and Correcting Effective Target Shift in Online Learning Authors: Ziyan Li, Naoki Hiratani Published: 2026-05-08 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2605.07886v1 Online learning from a stream of data is a defining feature of intelligence, yet modern machine learning systems often struggle in this setting, especially under distributional shift. To understand its basic properties, we study the relationship between online and offline learning in the context of kernel regression. We derive a closed-form expression for the function learned by online kernel regression, revealing that online kernel regression is equivalent to offline regression with shifted, inaccurate target outputs. Conversely, we show that by compensating for this effective shift in the teaching signal through target correction, online kernel-based learning can provably learn the same predictor as its offline counterpart. We derive both a closed-form expression for this target correction and an iterative form that can be applied sequentially. Applying this framework to image classification tasks on CIFAR-10 and CORe50, we show that online stochastic gradient descent with iteratively corrected targets outperforms learning with the true targets in continual learning settings. This work therefore provides a basic framework for analyzing and improving online learning in non-stationary environments. --- ### Beyond "I cannot fulfill this request": Alleviating Rigid Rejection in LLMs via Label Enhancement Authors: Ying Zhang, Congyu Qiao, Xin Geng, Ning Xu Published: 2026-05-08 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.07883v1 Large Language Models (LLMs) rely on safety alignment to obey safe requests while refusing harmful ones. However, traditional refusal mechanisms often lead to "rigid rejection," where a general template (e.g., "I cannot fulfill this request") indiscriminately triggers refusals and severely undermines the naturalness of interactions between humans and LLMs. To address this issue, LANCE is proposed in this paper to ensure safe yet flexible and natural responses via label enhancement. Specifically, LANCE employs variational inference to perform label enhancement, predicting a continuous distribution across multiple rejection categories. These fine-grained rejection distributions provide multi-way textual gradients for a refinement model to neutralize the hazardous elements in the prompt, so that the LLMs could generate safe responses that avoid rigid rejections while preserving the naturalness of interactions. Experiments demonstrate that LANCE significantly alleviates the rigid rejection problem while maintaining high security standards, significantly outperforming existing baseline models in terms of helpfulness and naturalness of responses. --- ### Black-box model classification under the discriminative factorization Authors: Hayden Helm, Merrick Ohata, Carey Priebe Published: 2026-05-08 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2605.07878v1 Access to modern generative systems is often restricted to querying an API (the ``black-box" setting) and many properties of the system are unknown to the user at inference time. While recent work has shown that low-dimensional representations of models based on the relationship between their embedded responses to a set of queries are useful for inferring model-level properties, the quality of these representations is highly sensitive to the query set. We introduce the \emph{discriminative factorization} to distinguish between high- and low-quality query sets in the context of black-box model-level classification. Under this framework, the probability of chance-level classification decays exponentially in the query budget. On three auditing tasks, estimated factorization parameters predict the empirical performance decay rate. We conclude by showing that query sets selected using the estimated discriminative field reproduce the empirical ordering of oracle query sets. --- ### Video Understanding Reward Modeling: A Robust Benchmark and Performant Reward Models Authors: Yuancheng Wei, Linli Yao, Lei Li, Haojie Zhang, Hao Zhou, Fandong Meng, Xu Sun Published: 2026-05-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.07872v1 Multimodal reward models have advanced substantially in text and image domains, yet progress in video understanding reward modeling remains severely limited by the lack of robust evaluation benchmarks and high-quality preference data. To address this, we propose a unified framework spanning benchmark design, data construction, and reward model training. We introduce Video Understanding Reward Bench (VURB), a benchmark featuring 2,100 preference pairs with long chain-of-thought reasoning traces (averaging 1,143 tokens) and majority voting evaluation across general, long, and reasoning-oriented video tasks. We further construct Video Understanding Preference Dataset (VUP-35K) via a fully automated pipeline, providing large-scale high-quality supervision for video reward training. Building on the data, we train VideoDRM and VideoGRM, a discriminative and a generative reward model, both achieving state-of-the-art performance on VURB and VideoRewardBench. Further analysis confirms that VUP-35K enhances both reward performance and model reasoning capability, while VideoDRM and VideoGRM yield significant gains under best-of-$N$ test-time scaling. --- ### Spectral Dynamics in Deep Networks: Feature Learning, Outlier Escape, and Learning Rate Transfer Authors: Clarissa Lauditi, Cengiz Pehlevan, Blake Bordelon Published: 2026-05-08 Categories: cond-mat.dis-nn, cs.AI, stat.ML arXiv: https://arxiv.org/abs/2605.07870v2 We study the evolution of hidden-weight spectra in wide neural networks trained by (stochastic) gradient descent. We develop a two-level dynamical mean-field theory (DMFT) that jointly tracks bulk and outlier spectral dynamics for spiked ensembles whose spike directions remain statistically dependent on the random bulk. We apply this framework to two settings: (1) infinite-width nonlinear networks in mean-field/$μ$P scaling and (2) deep linear networks in the proportional high-dimensional limit, where width, input dimension, and sample size diverge with fixed ratios. Our theory predicts how outliers evolve with training time, width, output scale, and initialization variance. In deep linear networks, $μ$P yields width-consistent outlier dynamics and hyperparameter transfer, including width-stable growth of the leading NTK mode toward the edge of stability (EoS). In contrast, NTK parameterization exhibits strongly width-dependent outlier dynamics, despite converging to a stable large-width limit. We show that this bulk+outlier picture is descriptive of simple tasks with small output channels, but that tasks involving large numbers of outputs (ImageNet classification or GPT language modeling) are better described by a restructuring of the spectral bulk. We develop a toy model with extensive output channels that recapitulates this phenomenon and show that edge of the spectrum still converges for sufficiently wide networks. --- ### KL for a KL: On-Policy Distillation with Control Variate Baseline Authors: Minjae Oh, Sangjun Song, Gyubin Choi, Yunho Choi, Yohan Jo Published: 2026-05-08 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2605.07865v1 On-Policy Distillation (OPD) has emerged as a dominant post-training paradigm for large language models, especially for reasoning domains. However, OPD remains unstable in practice due to the high gradient variance of its single-sample Monte Carlo estimator, and recipes for stable training are still immature. We propose vOPD (On-Policy Distillation with a control variate baseline), which casts OPD as policy-gradient RL and stabilizes it by introducing a control variate baseline-canonically a value function -- from the RL literature. We show that the OPD value function admits a closed form as the per-token negative reverse KL divergence between the student and the teacher, available directly from the already-computed forward pass with no additional critic or inference. Existing stabilization methods either compute the full token-level reverse KL over the entire vocabulary, adding significant overhead, or restrict it to a top-k support, biasing the objective. vOPD instead preserves the lightweight single-sample estimator, subtracting the value function as a detached baseline to keep the gradient unbiased while reducing variance. Furthermore, we show that a top-k approximation of the baseline further lowers cost without compromising performance. Across mathematical and scientific reasoning benchmarks, vOPD consistently outperforms vanilla OPD and matches the most expensive full-vocabulary baseline, offering an efficient stabilization of On-Policy Distillation through principled RL variance reduction. --- ### ADKO: Agentic Decentralized Knowledge Optimization Authors: Lucas Nerone Rillo, Zhanhong Jiang, Nastaran Saadati, Aditya Balu, Baskar Ganapathysubramanian, Chinmay Hegde, Soumik Sarkar Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07863v1 We present Agentic Decentralized Knowledge Optimization (ADKO), a framework for collaborative black-box optimization across autonomous agents that achieves sample efficiency, privacy preservation, heterogeneous-objective handling, and communication efficiency. Each agent maintains a private Gaussian Process (GP) surrogate trained on local data and communicates only through knowledge tokens-compact, lossy summaries containing directional signals, advantage scores, and optional language-model (LM) insights-without sharing raw data or model parameters. ADKO unifies GP-Upper Confidence Bound (GP-UCB), parallel Bayesian optimization, decentralized learning, and LM-guided discovery. We provide the first formal analysis of dual information loss: token compression, quantified via mutual-information-based fidelity, and LM approximation error, decomposed into bias and stochastic noise. Our main result shows cumulative regret decomposes into GP error, LM bias, LM noise, and compression loss, with necessary and sufficient conditions for sublinear regret. We also propose fidelity-aware token pruning to preserve high-information tokens under memory budget. Experiments on neural architecture search and scientific discovery validate the theory and show consistent improvements over strong baselines. --- ### From Synthetic to Real: Toward Identity-Consistent Makeup Transfer with Synthetic and Real Data Authors: Yue Yu, Jiayu Wang, Jiajia Shi, Zhiyu Tan, Hao Li, Jingjing Chen Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07861v2 Makeup transfer aims to apply the makeup style of a reference portrait to a source portrait while preserving identity and background. Early methods formulate this task as unsupervised image-to-image translation, relying on surrogate objectives and often yielding limited performance. Recent diffusion- and flow-based approaches instead exploit synthetic data for supervised training, leading to significant improvements. However, these methods still face two critical challenges: synthetic supervision frequently fails to faithfully preserve identity, and the domain gap between synthetic and real data limits generalization, resulting in degraded performance in complex real-world scenarios. To address these issues, this paper first proposes ConsistentBeauty, a novel data curation pipeline that ensures makeup fidelity and strict identity consistency within the synthesized data. Second, we propose RealBeauty, a synthetic-to-real post-training framework. Beyond supervised learning on curated synthetic data, we further adapt the model to real-world scenarios through reinforcement learning and design novel verifiable rewards tailored to the makeup transfer task. It allows the model to further benefit from real makeup patterns beyond synthetic supervision. In addition, we establish a new diverse benchmark for makeup transfer, covering a wide range of skin tones, ages, genders, poses, and makeup styles, thereby enabling a more comprehensive evaluation of model performance under diverse real-world conditions. Extensive experiments show that our method achieves state-of-the-art performance on multiple benchmarks and demonstrates clear advantages in identity preservation and performance on complex real-world cases. --- ### On the Tradeoffs of On-Device Generative Models in Federated Predictive Maintenance Systems Authors: Usevalad Milasheuski, Piero Baraldi, Enrico Zio, Stefano Savazzi Published: 2026-05-08 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.07860v1 Federated Learning (FL) has emerged as a promising paradigm for preserving client data ownership and control over distributed Internet of Things (IoT) environments. While discriminative models dominate most FL use cases, recent advances in generative models -- such as Variational Autoencoders (VAE), Generative Adversarial Networks (GAN), and Diffusion Models (DM) -- offer new opportunities for unsupervised anomaly detection in time series analysis, with relevant applications in predictive maintenance (PdM) in critical industrial infrastructures. In this work, we present a comprehensive analysis of VAEs, GANs, and DMs in the context of federated PdM. We analyze their performance and communication overhead under both full and partial federation setups, where only subsets of model components are shared. Building on this analysis, the paper proposes a novel taxonomy for federated generative models that formalizes partial component sharing as a principled mechanism for model personalization. Our experiments over a real-world time series dataset reveal distinct trade-offs in model utility, stability, and scalability, especially in heterogeneous and bandwidth-constrained FL settings. For the evaluated GAN-based configurations, full federation improves training stability relative to independent local training, although the model remains less robust than the VAE- and DDPM-based alternatives. For DMs, however, partial federation -- especially decoder sharing -- can outperform full federation in bandwidth-constrained, non-IID settings. --- ### EyeCue: Driver Cognitive Distraction Detection via Gaze-Empowered Egocentric Video Understanding Authors: Lang Zhang, JinYi Yoon, Matthew Corbett, Abhijit Sarkar, Bo Ji Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07859v1 Driver cognitive distraction is a major cause of road collisions and remains difficult to detect. Unlike manual or visual distraction, cognitive distraction is diverted by thoughts unrelated to driving, even when the driver appears visually attentive and exhibits no explicit physical movements. In this work, we propose EyeCue, a gaze-empowered egocentric video understanding framework, to detect driver cognitive distraction. A key insight is that cognitive distraction manifests in the interaction between eye gaze and visual context. To capture this interaction, EyeCue integrates eye gaze with egocentric video to enable context-aware modeling of the driver's attention over time. Furthermore, to tackle the limited scale and diversity of existing datasets, we introduce CogDrive, a comprehensive multi-scenario dataset that augments four existing driving datasets with cognitive distraction annotations. Through extensive evaluations on CogDrive, we show that EyeCue achieves the highest accuracy of 74.38%, outperforming 11 baselines from 6 model families by over 7%. Notably, EyeCue can achieve an accuracy of over 70% across various driving scenarios (different road types, times of day, and weather conditions) with strong generalizability. These results highlight the importance of modeling gaze-context interactions and the effectiveness of cross-modal interaction modeling for multimodal cognitive distraction detection. Our codes and CogDrive dataset resources are available at https://github.com/langzhang2000/EyeCue. --- ### Actor-Critic Algorithm for Dynamic Expectile and CVaR Authors: Yudong Luo, Erick Delage Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07857v1 Optimizing dynamic risk with stochastic policies is challenging in both policy updates and value learning. The former typically requires transition perturbation, while the latter may rely on model-based approaches. To address these challenges, we propose a surrogate policy gradient without transition perturbation under softmax policy parameterization. We further develop model-free value learning methods for dynamic expectile and conditional value-at-risk by leveraging elicitability. Finally, inspired by Expected SARSA and Expected Policy Gradient, a model-free off-policy actor-critic algorithm is constructed. Empirical results in domains with verifiable risk-averse behavior show that our algorithm can learn risk-averse policy and consistently outperforms other existing methods. --- ### MatryoshkaLoRA: Learning Accurate Hierarchical Low-Rank Representations for LLM Fine-Tuning Authors: Ionut-Vlad Modoranu, Mher Safaryan, Dan Alistarh Published: 2026-05-08 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.07850v1 With the rise in scale for deep learning models to billions of parameters, the computational cost of fine-tuning remains a significant barrier to deployment. While Low-Rank Adaptation (LoRA) has become the standard for parameter-efficient fine-tuning, the need to set a predefined, static rank $r$ requires exhaustive grid searches to balance efficiency and performance. Existing rank-adaptive solutions such as DyLoRA mitigate this by sampling ranks during the training from a predefined distribution. However, they often yield sub-optimal results at higher ranks due to lack of consistent gradient signals across the full hierarchy of ranks, thus making these methods data-inefficient. In this paper, we propose MatryoshkaLoRA, a general, Matryoshka-inspired training framework for LoRA that learns accurate hierarchical low-rank representations by inserting a fixed, carefully crafted diagonal matrix $P$ between the existing LoRA adapters to scale their sub-ranks accordingly. By introducing this simple modification, our general framework recovers LoRA and DyLoRA only by changing $P$ and ensures all sub-ranks embed the available gradient information efficiently. Our MatryoshkaLoRA supports dynamic rank selection with minimal degradation in accuracy. We further propose Area Under the Rank Accuracy Curve (AURAC), a metric that consistently evaluates the performance of hierarchical low-rank adapters. Our results demonstrate that MatryoshkaLoRA learns more accurate hierarchical low-rank representations than prior rank-adaptive approaches and achieves superior accuracy-performance trade-offs across ranks on the evaluated datasets. Our code is available at https://github.com/IST-DASLab/MatryoshkaLoRA. --- ### Measuring and Mitigating the Distributional Gap Between Real and Simulated User Behaviors Authors: Shuhaib Mehri, Philippe Laban, Sumuk Shashidhar, Marwa Abdulhai, Sergey Levine, Michel Galley, Dilek Hakkani-Tür Published: 2026-05-08 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.07847v1 As user simulators are increasingly used for interactive training and evaluation of AI assistants, it is essential that they represent the diverse behaviors of real users. While existing works train user simulators to generate human-like responses, whether they capture the broad and heterogeneous distribution of real user behaviors remains an open question. In this work, we introduce a method to measure the distributional gap between real and simulated user behaviors, validated through a human study and ablations. Given a dataset of real and simulated conversations, our method extracts representations of user behavior from each conversation, quantizes them into discrete distributions via clustering, then computes divergence metrics. We provide the first systematic evaluation of 24 LLM-based user simulators on coding and writing tasks, and reveal a large distributional gap from real users that varies across model families, scales, and behavioral facets. Pairwise comparisons show that most simulators behave similarly, while a few stand apart. Combining behaviorally complementary simulators brings the resulting distribution closer to real users compared to either simulator on its own. Finally, a TF-IDF analysis of the clusters surfaces interpretable patterns of behaviors that simulators capture, miss, and hallucinate. --- ### BRIDGE: Background Routing and Isolated Discrete Gating for Coarse-Mask Local Editing Authors: Peilin Xiong, Honghui Yuan, Junwen Chen, Keiji Yanai Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07846v3 Coarse-mask local image editing asks a model to modify a user-indicated region while preserving the surrounding scene. In practice, however, rough masks often become unintended shape priors: instead of serving as flexible edit support, the mask can pull the generated object toward its accidental boundary. We study this failure as mask-shape bias and frame the task through a Two-Zone Constraint, where the background should remain stable while the editable region should follow the instruction without being forced to inherit the mask contour. BRIDGE addresses this setting by keeping masks outside the DiT backbone for support construction and blending, avoiding DiT-internal mask injection and copied control branches. It uses BridgePath generation, where a Main Path preserves background context and a Subject Path generates editable content from independent noise. Motivated by a diagnostic Qwen-Image experiment showing that positional embeddings and attention connectivity regulate which image context visual tokens reuse, BRIDGE introduces a learnable Discrete Geometric Gate for token-level positional-embedding routing. This gate lets subject tokens borrow background-anchored coordinates near fusion regions or keep subject-centric coordinates for geometric freedom. We evaluate BRIDGE on BRIDGE-Bench, MagicBrush, and ICE-Bench. On BRIDGE-Bench, BRIDGE improves Local SigLIP2-T from 0.262 with FLUX.1-Fill and 0.390 with ACE++ to 0.503, with parallel gains in local DINO and DreamSim. Zero-shot results on MagicBrush and ICE-Bench further indicate competitive alignment and source preservation beyond the curated benchmark, while the added routing module remains compact at 13.31M parameters compared with ControlNet-style copied branches. --- ### BRIDGE: Background Routing and Isolated Discrete Gating for Coarse-Mask Local Editing Authors: Peilin Xiong, Honghui Yuan, Junwen Chen, Keiji Yanai Published: 2026-05-08 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.07846v2 Coarse-mask local image editing asks a model to modify a user-indicated region while preserving the surrounding scene. In practice, however, rough masks often become unintended shape priors: instead of serving as flexible edit support, the mask can pull the generated object toward its accidental boundary. We study this failure as mask-shape bias and frame the task through a Two-Zone Constraint, where the background should remain stable while the editable region should follow the instruction without being forced to inherit the mask contour. BRIDGE addresses this setting by keeping masks outside the DiT backbone for support construction and blending, avoiding DiT-internal mask injection and copied control branches. It uses BridgePath generation, where a Main Path preserves background context and a Subject Path generates editable content from independent noise. Motivated by a diagnostic Qwen-Image experiment showing that positional embeddings and attention connectivity regulate which image context visual tokens reuse, BRIDGE introduces a learnable Discrete Geometric Gate for token-level positional-embedding routing. This gate lets subject tokens borrow background-anchored coordinates near fusion regions or keep subject-centric coordinates for geometric freedom. We evaluate BRIDGE on BRIDGE-Bench, MagicBrush, and ICE-Bench. On BRIDGE-Bench, BRIDGE improves Local SigLIP2-T from 0.262 with FLUX.1-Fill and 0.390 with ACE++ to 0.503, with parallel gains in local DINO and DreamSim. Zero-shot results on MagicBrush and ICE-Bench further indicate competitive alignment and source preservation beyond the curated benchmark, while the added routing module remains compact at 13.31M parameters compared with ControlNet-style copied branches. --- ### Distributional simplicity bias and effective convexity in Energy Based Models Authors: Aurélien Decelle, Alfonso de Jesús Navas Gómez, Beatriz Seoane Published: 2026-05-08 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.07844v1 Energy-based learning is a powerful framework for generative modelling, but its training is inherently non-convex, leading potentially to sensitivity to initialisation, poor local optima, and unstable gradient dynamics. We present a dynamical analysis of energy-based learning through the lens of the effective model, which can be interpreted as either a generalised Ising model with higher-order interactions or the Fourier expansion of the energy. Under sufficient expressivity, we show that the gradient flow induced by learning strictly positive distributions over binary variables admits two types of fixed points: data-consistent points, which exactly reproduce the target distribution, and spurious points, which satisfy stationarity without matching the target distribution. Around data-consistent points, we show that perturbations are either stable or neutral, with neutral directions leaving the effective model invariant. Finally, we show that gradient dynamics induce a hierarchy in which lower-order interactions are learned before higher-order ones. This provides a mechanistic explanation for the distributional simplicity bias and clarifies why fixed points that are not data-consistent at low orders are not observed in practice. --- ### \mathsf{VISTA}: Decentralized Machine Learning in Adversary Dominated Environments Authors: Hanzaleh Akbari Nodehi, Parsa Moradi, Soheil Mohajer, Mohammad Ali Maddah-Ali Published: 2026-05-08 Categories: cs.LG, cs.AI, cs.DC arXiv: https://arxiv.org/abs/2605.07841v1 Decentralized machine learning often relies on outsourcing computations, such as gradient evaluations, to untrusted worker nodes. Existing robust aggregation methods can mitigate malicious behavior under honest-majority assumptions, but may fail when adversaries control a majority of the workers. We study this adversary-dominated setting through an incentive-oriented framework in which reports are accepted and rewarded only when they are mutually consistent up to a threshold. This turns the adversary from a pure saboteur into a rational agent that trades off increasing estimation error against the risk of rejection and loss of reward. We consider iterative optimization under this model. Unlike one-shot computation, iterative learning requires long-horizon decisions: permissive acceptance rules enable faster early progress but admit more adversarial corruption, while strict rules improve estimation accuracy but cause frequent rejections. We propose \mathsf{VISTA}, an adaptive algorithm that tunes the acceptance threshold using the optimization history. Numerical results show that \mathsf{VISTA} improves convergence over static thresholds. We also provide a rigorous convergence analysis showing that, with suitable incentive-aware adaptation, adversary-dominated decentralized learning can retain the asymptotic convergence behavior of standard SGD without relying on an honest majority. ---