探索全球前沿学术脉络

01.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2603.12261

The Latent Color Subspace: Emergent Order in High-Dimensional Chaos

作者:

Mateusz Pach↗Jessica Bader↗Quentin Bouniot↗Serge Belongie↗Zeynep Akata↗

Text-to-image generation models have advanced rapidly, yet achieving fine-grained control over generated images remains difficult, largely due to limited understanding of how semantic information is encoded. We develop an interpretation of the color representation in the Variational Autoencoder latent space of FLUX.1 [Dev], revealing a structure reflecting Hue, Saturation, and Lightness. We verify our Latent Color Subspace (LCS) interpretation by demonstrating that it can both predict and explicitly control color, introducing a fully training-free method in FLUX based solely on closed-form latent-space manipulation. Code is available at https://github.com/ExplainableML/LCS.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.CV) 2026-06-18 DOI: arXiv:2606.18824

Where Will They Go? Modelling Multimodal Pedestrian Manoeuvres from Ego-centric Videos

作者:

Yuxuan Xie↗Nicolas Pugeault↗Chongfeng Wei↗Hubert P. H. Shum↗Edmond S. L. Ho↗

Pedestrian trajectory prediction from an ego-centric camera is challenging since it depends on complex interactions with vehicles and scene context, as well as the intention of the pedestrian. By modelling correlation and intent from the historical and future trajectories of the pedestrian, it will usually result in a multimodal (i.e. multiple modes) distribution. Existing stochastic predictors often sample multiple futures from a single unimodal distribution, which can yield sub-optimal 'mixed-mode' trajectories that lie between distinct motion patterns and become implausible in real scenes. In this paper, we propose MMPM, a mode-aware framework that separately models future trajectory distributions into semantically meaningful modes based on the pedestrian's crossing behavior. MMPM consists of two modules: behavior-aware Pedestrian Interaction Module (PIM) that jointly captures pedestrian-vehicle and pedestrian-environment interactions by introducing gaze, head and hand gesture, and a CVAE-based Mode-aware Trajectory Predictor (MTP) module to model the future trajectory distributions on two modes, crossing and non-crossing the road, separately. A query-based decoder further enforces mode consistency during decoding. Experiments on PIE and JAAD datasets show that our method surpasses state-of-the-art baselines. Our proposed MTP is model-agnostic, which can be integrated into existing frameworks such as BiTrap-NP and SGNet-ED to further improve future trajectory prediction performance. We additionally introduce a data-driven validation protocol that matches predictions to spatio-temporally consistent ground-truth trajectories, demonstrating improved frame-wise displacement errors over previous work.

阅读与讨论 → 访问原文 →

03.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.09365

Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory

作者:

Haoran Sun↗Wenjie Li↗Yujie Zhang↗Zekai Lin↗Fanrui Zhang↗Kaitao Chen↗Xingqi He↗Yichen Li↗Mianxin Liu↗Lei Liu↗Yankai Jiang↗

Medical agent systems are increasingly expected to support interactive clinical decision making rather than only static question answering. In such settings, effective agents must reuse prior experience across evolving cases, yet existing memory mechanisms often retain raw historical traces that are redundant, noisy, and difficult to govern. More importantly, they rarely distinguish which memories are truly useful for future reasoning. This limits their ability to accumulate compact and reliable experience for long-horizon clinical reasoning. To close this gap, we propose SkeMex, a post-deployment self-evolution framework that improves medical agents through a skill-based memory without updating model weights. SkeMex distills informative interaction trajectories into structured skills that encode reusable procedural knowledge, and organizes them into a multi-branch repository spanning general, task-specific, and action-level experience. To determine which memories should be reused and retained, SkeMex estimates context-dependent utility from environment feedback and uses it to guide value-aware retrieval and repository governance. A closed-loop ``Read–Write–Assess–Govern" lifecycle further supports continual evolution by writing new skills, updating utilities, promoting useful memories, and removing harmful entries. Experiments across diverse clinical tasks show that SkeMex consistently outperforms representative memory-based agents in both offline and online settings. It also generalizes across model backbones and supports transferable skill memory. All data and code will be released publicly.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.15629

XPASS-Vis: A Dataset for Cross-Domain Personalized Image Aesthetic Assessment

作者:

Takato Hayashi↗Hiroaki Takahara↗Candy Olivia Mawalim↗Hiromi Narimatsu↗Akisato Kimura↗Shiro Kumano↗Shogo Okada↗

Personalized image aesthetic assessment (PIAA) seeks to model, at the individual level, the subjective nature of aesthetic judgments toward artworks and photographs. Aesthetic preference is known to be both deeply personal and partially consistent across visual domains. Yet existing PIAA datasets and methods are largely confined to a single domain, or provide too few samples per annotator within each domain to enable personalization across domains. Consequently, the cross-domain generalization of personalized aesthetic preferences remains largely unexplored. To address this gap, we introduce XPASS-Vis, the first dataset explicitly designed for cross-domain PIAA. XPASS-Vis comprises 6,526 stimuli from three visual domains – art, fashion, and landscape – rated by 129 annotators, yielding 87,836 user-stimulus interactions, each annotated with an overall aesthetic score and nine aesthetic-emotion ratings. Notably, each annotator rated more than 200 stimuli per domain, providing sufficient per-domain coverage to support personalization both within and across domains. Moreover, we establish baseline models for cross-domain PIAA under unsupervised domain adaptation (UDA), where a model trained on a labeled source domain is transferred to an unlabeled target domain. A systematic evaluation of representative UDA approaches shows that the best-performing method recovers approximately 60\% (Spearman's $\rho$ = .28) of the supervised upper bound under a fully unsupervised setting. This provides encouraging evidence that personalized aesthetic preferences are, to a meaningful extent, transferable across visual domains. At the same time, a substantial gap remains, highlighting the need for PIAA-specific adaptation strategies. XPASS-Vis and the accompanying baselines provide a foundation for future research on cross-domain PIAA. All datasets and code will be made publicly available upon acceptance.

阅读与讨论 → 访问原文 →

05.

arXiv (CS.AI) 2026-06-18 DOI: arXiv:2606.19004

Spotlight: Synergizing Seed Exploration and Spot GPUs for DiT RL Post-Training

作者:

Ruiqi Lai↗Dakai An↗Wei Gao↗Ju Huang↗Siran Yang↗Jiamang Wang↗Lin Qu↗Dmitrii Ustiugov↗Wei Wang↗

arXiv:2606.19004v1 Announce Type: cross Abstract: Reinforcement learning (RL) post-training of Diffusion Transformers (DiTs) is prohibitively expensive, requiring thousands of high-end GPUs. Existing works explore two directions to reduce cost: seed exploration improves training convergence by selecting high-contrast samples, yet adds compute to the critical path; spot GPUs offer 69–77\% lower cost, yet sit idle during training because DiT rollouts finish nearly simultaneously, which prevents LLM-style pipelining of rollout with training. Spot preemptions further break Sequence Parallelism (SP) groups, fragmenting GPU topology. We present Spotlight, the first system that harvests spot GPUs for DiT RL post-training. Spotlight rests on two key insights we devise: (1)~we show that exploration can tolerate stale model weights because exploration that uses the model weights from the previous iteration preserves the relative ranking of random seeds, allowing exploration to run on idle spot GPUs during training. (2)~SP reconfiguration can reuse on-node state, reducing group recovery from minutes to sub-second launches. Built on these insights, Spotlight introduces three techniques: a bandit-based exploration planner that maximizes reward variance within the training time budget, elastic sequence parallelism that reconfigures SP groups on the fly via persistent schedulers and intra-node weight copying, and a preemption-aware pull-based request scheduler that balances load and commits in-flight state upon preemption. We implement Spotlight on the open-source RL platform ROLL and evaluate it on Qwen-Image post-training. Spotlight reaches the same target validation score $4\times$ faster than baselines, reducing total cost by $1.4$-$6.4\times$ while achieving superior image quality on DeepSeek-OCR and Geneval datasets with resolution $512\times512$ and $1280\times1280$.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.16567

TNODEV: Toolbox for Neural ODE Verification

作者:

Abdelrahman Sayed Sayed↗Pierre-Jean Meyer↗Mohamed Ghazel↗

arXiv:2606.16567v1 Announce Type: new Abstract: Neural ordinary differential equations (neural ODE) have started to appear in safety critical settings such as continuous-time controllers for cyber-physical systems and classifiers integrated into automated decision pipelines, raising the question of whether their behavior can be formally verified. Existing tools dedicated to neural ODE provide only a single reachability call without iterative input set refinement, limiting the precision of their verdicts to whatever one reachability call can deliver. We present TNODEV, the first sound formal verifier for neural ODE that integrates a falsification checker, a fast interval-based reachability backend based on continuous-time mixed monotonicity, a verification and refinement loop with three input-set splitting heuristics, and a parallel scheduler in a single end-to-end pipeline. TNODEV supports safe-set inclusion verification on pure neural ODE, neural ODE in closed loop with a neural network controller and general neural ODE (GNODE), with the safe set specified either as an interval or as the half-space intersection induced by a target classification label. We evaluate TNODEV on a range of benchmarks across safe-set inclusion and classification-robustness properties, including a direct reachability comparison against NNV~2.0 and CORA and a verification comparison against NNV2.0 on MNIST general neural ODE classifiers.

阅读与讨论 → 访问原文 →

07.

arXiv (CS.CL) 2026-06-18 DOI: arXiv:2510.04120

Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing

作者:

Fengying Ye↗Shanshan Wang↗Lidia S. Chao↗Derek F. Wong↗

Large language models (LLMs) achieve strong performance on metaphor detection and interpretation tasks, yet it remains unclear what such behavioral success reveals about metaphor processing. We present a diagnostic analysis that examines the limits of behavioral evidence by probing three complementary dimensions: semantic attribute alignment, lexical invariance, and syntactic sensitivity. Using geometric probing, we assess whether model-generated interpretations align with reference semantic attributes; through context-varying substitution, we analyze the stability of lexical associations between metaphorical and literal expressions; and via controlled syntactic perturbations, we examine sensitivity in metaphor detection. Our analysis reveals that LLM-generated interpretations can exhibit semantic drift relative to reference attributes; stable lexical anchors persist across contextual conditions, potentially supporting conventional metaphors while biasing novel metaphors requiring contextual integration; and detection performance is sensitive to syntactic irregularities. These findings suggest that strong behavioral performance may reflect heterogeneous underlying signals, highlighting the need for caution when interpreting metaphor benchmarks as evidence of robust, integrated semantic understanding.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.18537

Do as the Romans Do: Learning Universal Behaviors from Heterogeneous Agents

作者:

Caleb Chang↗Davin Win Kyi↗Natasha Jaques↗Karen Leung↗

arXiv:2606.18537v1 Announce Type: new Abstract: Humans often acquire new skills by observing others, since observed behaviors implicitly reveal how to act in an environment. However, observations drawn from a heterogeneous population introduce conflicting behavioral signals, making it difficult to determine which behaviors are worth imitating. We address this challenge with General Reward Inference and Disentanglement (GRID), a social learning method that extracts universally useful behaviors from a heterogeneous population of demonstrators pursuing different goals. GRID decomposes per-agent reward functions into a general reward, capturing behaviors shared across all agents, and specific rewards, capturing individual preferences and objectives. Training exclusively on the general reward provides a new paradigm of generalist pretraining. It yields a generalist agent that internalizes universal environmental competencies, such as safety and basic task proficiency, without the mode-averaging bias that afflicts standard learning from demonstration techniques. This generalist serves as a superior prior for fine-tuning to downstream tasks, including preferences unseen during training. Experiments across a synthetic basis function decomposition, multi-agent Craftax, and a continuous autonomous driving simulator (Highway-Env) confirm that GRID successfully disentangles reward structure in a semantically meaningful way, outperforms standard learning from demonstration baselines, and enables more efficient and stable specialization.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2512.16415

CountZES: Counting via Zero-Shot Exemplar Selection

作者:

Muhammad Ibraheem Siddiqui↗Muhammad Haris Khan↗

Object counting in complex scenes is particularly challenging in the zero-shot (ZS) setting, where instances of unseen categories are counted using only a class name. Existing ZS counting methods that infer exemplars from text often rely on off-the-shelf open-vocabulary detectors (OVDs), which in dense scenes suffer from semantic noise, appearance variability, and multi-instance proposals. Alternatively, random image-patch sampling is employed, which fails to accurately delineate object instances. Since counting is sensitive to exemplar quality, such selection strategies often yield poorly representative exemplars, leading to inaccurate count estimation. To address these issues, we propose CountZES, an inference-only approach for object counting via ZS exemplar selection. CountZES discovers diverse exemplars through three synergistic stages: Detection-Anchored Exemplar (DAE), Density-Guided Exemplar (DGE), and Feature-Consensus Exemplar (FCE). DAE refines OVD detections to isolate precise single-instance exemplars. DGE introduces a density-driven, self-supervised paradigm to identify statistically consistent and semantically compact exemplars, while FCE reinforces visual coherence through feature-space clustering. Together, these stages yield a complementary exemplar set that balances textual grounding, count consistency, and feature representativeness. Experiments on diverse datasets demonstrate CountZES superior performance among ZOC methods while generalizing effectively across domains.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.10820

K-Forcing: Joint Next-K-Token Decoding via Push-Forward Language Modeling

作者:

Zhiwei Tang↗Yuanyu He↗Yizheng Han↗Wangbo Zhao↗Jiasheng Tang↗Fan Wang↗Bohan Zhuang↗

Autoregressive (AR) language modeling is the dominant paradigm for text generation, yet its sequential token-by-token decoding makes inference memory-bound and inefficient. Existing acceleration approaches, such as speculative decoding and diffusion language models, can yield speedups under certain conditions but do not directly address high-load batch serving–the scenario most critical for industrial-scale deployment. We introduce K-Forcing, a push-forward language modeling paradigm for joint next-k-token decoding. K-Forcing distills an existing AR model into a conditional push-forward mapping–one that transforms independent uniform noise variables into a joint sample of multiple future tokens in a single forward pass. This design preserves fixed-length outputs, reuses the AR teacher backbone, and remains compatible with standard AR serving infrastructure. We train this mapping via progressive self-forcing distillation, which gradually expands the prediction window while enabling the student to closely match the sequence distribution of the AR teacher. We evaluate K-Forcing on LM1B and OpenWebText using a standard causal Transformer backbone. When aggressively configured to generate k = 4 tokens per forward pass, K-Forcing delivers approximately 2.4-3.5x speedup across different batch sizes, while incurring modest quality degradation relative to its AR teacher. As inference increasingly dominates the lifetime compute cost of modern LLMs, K-Forcing offers a promising route toward accelerating AR generation under real-world high-load deployment.

阅读与讨论 → 访问原文 →

11.

arXiv (quant-ph) 2026-06-11 DOI: arXiv:2603.11172

Integrable Massless and Massive Fermions

作者:

Zhao Zhang↗

arXiv:2603.11172v2 Announce Type: replace-cross Abstract: One-dimensional integrable fermions can be classified into massless and massive regimes, and the $R$-operator for the latter can be constructed from that of the former. Here, I define integrable massless fermions by the simultaneous satisfaction of the Yang-Baxter equation (YBE) and Shastry's decorated YBE (DYBE) by the $R$-matrix. This notion is strictly more general than Maassarani's `free-fermion algebra', yet more restrictive than the notion of free fermions in exactly solvable quantum models or in integrable two-dimensional classical vertex models dual to quantum spin chains. Within this framework, there emerge two archetypal mechanisms for opening a spectral gap and generating massive fermions: (i) breaking time-reversal symmetry by coupling to external field, and (ii) introducing time-reversal symmetric interactions. These paradigms are realized, respectively, in the XY chain in a longitudinal field and in the Hubbard model, both of which possess non-relativistic, bivariate $R$-matrices. Integrability conditions on local Hamiltonians for both massless and massive fermions are identified, and schematic procedures for uniquely determining their $R$-matrices are proposed.

阅读与讨论 → 访问原文 →

12.

arXiv (CS.AI) 2026-06-15 DOI: arXiv:2606.00947

Silent Failures in Federated Personalization of Foundation Models

作者:

YongKyung Oh↗Alex Bui↗

arXiv:2606.00947v2 Announce Type: replace-cross Abstract: Foundation models are increasingly personalized on decentralized private data through federated learning and are now deployed at scale under growing regulatory requirements for post-market monitoring. We argue that this convergence creates a distinct and under-recognized class of trustworthiness failures, which we term "Silent Failures." These include amplified bias, fairness collapse, and alignment erosion that may remain difficult to detect because federated learning's privacy constraints limit visibility into model behavior. A landscape analysis of existing benchmarks reveals a structural divide. Federated benchmarks evaluate system performance but provide limited insight into model behavior, whereas centralized trustworthiness benchmarks assess behavior but require model access incompatible with federated privacy. We introduce a taxonomy of six silent failure modes arising from the interaction of foundation model personalization, dataset shift, and core federated constraints. Our analysis shows that privacy-preserving training alone is insufficient for trustworthy deployment. We conclude with a research agenda for privacy-preserving behavioral evaluation and propose that silent failures become a standard diagnostic category for trustworthy federated artificial intelligence.

阅读与讨论 → 访问原文 →

13.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2606.17572

When Dynamics Models Read the Wrong Time Steps: Label-Free Event Credit Re-Anchoring for Robust Global Readouts

作者:

Yifan Wang↗

arXiv:2606.17572v1 Announce Type: new Abstract: Learned dynamics models often answer global physical questions, such as fault severity or impact stiffness, by pooling a per-step feature sequence into one readout vector. This sequence-to-global interface creates an under-studied temporal credit problem: with only trajectory-level supervision, a model can predict accurately in training conditions while reading from abundant smooth correlates rather than the brief physical events that determine the target. We call this failure temporal credit dilution. It is not exposed by the training loss and is not removed by standard physics-informed residuals, because the error lies in where the global readout assigns functional credit. We introduce Credit-in-Event, an interface-level probe for measuring how much pooled credit lands on event steps, and prove in closed form that a pooled linear reader routes credit to a spurious background channel as the event fraction shrinks. We then propose CREST, a training-free and label-free readout that estimates a transient event core from learned features and re-anchors the pooled representation through event-versus-rest contrast. Across simulated gear and impact systems, recurrent and attention encoders, and public bearing vibration data, CREST reduces out-of-distribution error while restoring event credit. Ablations show that stable-step selection and receptive-field shrinking fail, confirming that the gain comes from event-core credit re-anchoring rather than a generic locality or stability prior.

阅读与讨论 → 访问原文 →

14.

arXiv (math.PR) 2026-06-12 DOI: arXiv:2606.13648

Stochastic dominations for FK percolation and sharp thinning thresholds for the Ising energy field

作者:

Paul Cahen↗Avelio Sep\'ulveda↗

arXiv:2606.13648v1 Announce Type: new Abstract: At first glance, one would imagine that the energy field of the Ising model, the set of edges whose endpoints share the same spin, is stochastically monotone as a function of the coupling constants. However, this is not generally the case. In this paper, we introduce two weaker notions of stochastic domination that make this result true: $p$–weak and $p$–weak$^\dagger$ domination. Both of these notions depend on a parameter $p$ and we find the optimal values $p$ and $p^\dagger$ so that these dominations hold. One of the key ingredient to obtain some of the results is a new stochastic domination relating FK percolations with different parameters $q,\tilde{q}\geq 1$ that is of independent interest.

阅读与讨论 → 访问原文 →

15.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2604.03496

Beyond Predefined Schemas: TRACE-KG for Context-Enriched Knowledge Graph Generation

作者:

Mohammad Sadeq Abolhasani↗Yang Ba↗Yixuan He↗Rong Pan↗

arXiv:2604.03496v2 Announce Type: replace Abstract: Knowledge graph generation typically relies either on predefined ontologies or on schema-free extraction. Ontology-driven pipelines enforce consistent typing but require costly schema design and maintenance, whereas schema-free methods often produce fragmented graphs with weak global organization, especially in long technical documents with dense, context-dependent information. We propose TRACE-KG (Text-dRiven schemA for Context-Enriched Knowledge Graphs), a framework that jointly constructs a context-enriched knowledge graph and an induced schema without assuming a predefined ontology. TRACE-KG captures conditional relations through structured qualifiers and organizes entities and relations using a data-driven schema that serves as a reusable semantic scaffold while preserving full traceability to the source evidence. Experiments show that TRACE-KG produces structurally coherent, traceable knowledge graphs and offers a practical alternative to both ontology-driven and schema-free construction pipelines.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.LG) 2026-06-11 DOI: arXiv:2606.11255

Bernstein-Schur Kernels: Random Features by Sketched Modulation and Radial Randomization

作者:

Taha Bouhsine↗

arXiv:2606.11255v1 Announce Type: new Abstract: Bernstein–Schur kernels are products of a finite-feature kernel (one with an explicit finite-dimensional feature map) and a completely monotone shift-invariant kernel: nonstationary kernels that fall between the shift-invariant and dot-product templates random features usually exploit, so in general neither Bochner sampling nor polynomial sketching applies to the full kernel directly. We give one random-feature construction for the whole class that randomizes both factors: it sketches the finite modulation and randomizes the completely monotone radial factor, sampling the latter's one-dimensional Bernstein–Widder scale and then applying Gaussian random Fourier features (whose frequency is still $d$-dimensional). The feature dimension is then $Dm$, set by the sketch size $m$ and the radial-draw count $D$, free of the $O(d^2)$ size of the exact modulation feature. Keeping the modulation \emph{exact is the analyzable limit ($m\to\infty$): there we prove unbiasedness, an exact variance for the recommended flat estimator, an expected matrix-Bernstein operator-norm bound (with a matching high-probability tail) controlled by the top eigenvalues of the kernel and modulation Gram matrices together with an intrinsic dimension rather than the crude $N\max_{ij}$ entrywise route, and a deterministic relative-spectral kernel-ridge stability result. By conditioning on the sketch, the doubly-randomized estimator inherits the same intrinsic-dimension operator-norm guarantee plus a single additive sketch term, tunable by $m$ independently of $D$. The motivating instance is the biased $yat$-kernel $k_{yat,b}(w,x)=(w^\top x+b)^2/(\|w-x\|^2+\varepsilon)$, $b\ge0$, whose family span contains the inverse-multiquadric kernel by finite differences in $b$; for it the radial mixture is the IMQ spectral sampler, and one frequency per scale is variance-optimal at a fixed radial-feature budget.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.CV) 2026-06-15 DOI: arXiv:2606.13861

Temporal Backtracking Search for Test-time Generative Video Reasoning

作者:

Sejoon Jun↗Zheng Ding↗Huangyuan Su↗Weirui Ye↗Yilun Du↗

While test-time scaling has revolutionized reasoning in large language models, generative video reasoning remains bottlenecked by a single-shot paradigm. We demonstrate that searching over denoising steps cannot rescue logically flawed rollouts because spatial trajectories commit early in the diffusion process. Root-level Best-of-N (BoN) sampling is similarly inefficient: reasoning errors cluster early in the temporal axis, and resampling blindly discards verified upstream progress. To unlock effective test-time scaling for video models, we introduce Temporal Backtracking Search (TBS), which shifts the search space to the temporal axis. TBS transforms video generation into an iterative generate-verify-restart loop via three core mechanisms: (1) variable-K conditioning to resume generation from arbitrary clean prefixes; (2) temporal process verification to localize failures and extract valid restart anchors; and (3) prefix-based search to reallocate compute toward extending correct trajectories rather than root resampling. Across algorithmic, navigation, and robotics domains, TBS Pareto-dominates matched-budget BoN. In a strict out-of-distribution setting where one-shot generation collapses (0.7% for BoN), TBS achieves 22.7%, with every solved episode stemming from a restarted branch. Ultimately, TBS reveals that the local reasoning competence of video models far exceeds what single-shot rollouts indicate, providing a scalable test-time framework to unlock it.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.15129

EyeMVP: OCT-Informed Fundus Representation Learning via Paired CFP–OCT Pretraining

作者:

Zhuo Deng↗Ruiheng Zhang↗Ziheng Zhang↗Weihao Gao↗Yitong Li↗Qian Wang↗Lei Shao↗Jiaoyue Dong↗Zhixi Zeng↗Lijian Fang↗Haibo Wang↗Xiaobin Lin↗…

Color fundus photography (CFP) is the mainstay for large-scale retinal screening, yet its diagnostic capacity is constrained by the lack of depth-resolved structural information. Optical coherence tomography (OCT) provides cross-sectional retinal anatomy, but is less accessible in population-level screening. Here, we present EyeMVP, a cross-modal retinal foundation model that uses paired CFP–OCT pretraining to learn OCT-informed CFP representations. EyeMVP is pretrained on 674,893 strict same-eye same-day paired CFP–OCT image triples from 112,642 patients across eight hospitals in China. The model uses cross-modal masked reconstruction to enrich CFP representations with OCT-associated supervision, while requiring only CFP images at inference. To accommodate the non-aligned imaging geometry between en-face CFP and cross-sectional OCT, EyeMVP combines source-constrained cross-attention with CFP-derived structural masks. Across 16 downstream tasks, including classification, segmentation, few-shot adaptation, and cross-modal retrieval, EyeMVP outperforms representative retinal foundation models and shows consistent gains on tasks involving macular and optic nerve structure. For CFP-challenging macular diseases, EyeMVP achieves an AUROC of 0.948 for macular edema (vs.~0.852 for EyeCLIP) and 0.825 for myopic macular schisis. In an exploratory reader study, EyeMVP exceeds junior and intermediate ophthalmologist groups but does not reach senior ophthalmologist performance on macular edema, while showing numerically higher balanced accuracy than all reader groups on myopic macular schisis. These results suggest that pixel-level cross-modal reconstruction can enrich CFP representations with OCT-associated supervision, providing a practical route toward stronger CFP-based retinal analysis in screening settings.

阅读与讨论 → 访问原文 →

19.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2510.01444

Dual-Uncertainty Guided Policy Learning for Multimodal Reasoning

作者:

Rui Liu↗Dian Yu↗Tong Zheng↗Runpeng Dai↗Zongxia Li↗Wenhao Yu↗Zhenwen Liang↗Linfeng Song↗Haitao Mi↗Pratap Tokekar↗Dong Yu↗

Reinforcement learning with verifiable rewards (RLVR) has advanced reasoning capabilities in multimodal large language models. However, existing methods typically treat visual inputs as deterministic, overlooking the perceptual ambiguity inherent to the visual modality. Consequently, they fail to distinguish whether a model's uncertainty stems from complex reasoning or ambiguous perception, preventing the targeted allocation of exploration or learning signals. To address this gap, we introduce DUPL, a dual-uncertainty guided policy learning approach for multimodal RLVR that quantifies and leverages both perceptual uncertainty (via symmetric KL divergence) and output uncertainty (via policy entropy) to guide policy updates. By establishing an uncertainty-driven feedback loop and employing a dynamic branch prioritization mechanism, DUPL recalibrates the policy advantage to focus learning on states with high perceptual or decisional ambiguity, enabling effective targeted exploration beyond passive data augmentation. Evaluated on diverse multimodal reasoning benchmarks spanning mathematical and general domains, DUPL achieves solid gains. It improves Qwen2.5-VL accuracy by up to $12.3%$ (3B) and $7.9%$ (7B), and Qwen3-VL-Instruct by up to $10.7%$ (4B) and $12.4%$ (8B), consistently outperforming GRPO, while seamlessly generalizing to alternative algorithms (DAPO, $+6.5%$ avg) and architectures (LLaVA-OneVision-1.5, $+4.7%$ avg). These results demonstrate that DUPL is an effective and generalizable approach for multimodal RLVR.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.AI) 2026-06-17 DOI: arXiv:2604.18701

Curiosity-Critic: Cumulative Prediction Error Improvement as a Tractable Intrinsic Reward for World Model Training

作者:

Vin Bhaskara↗Haicheng Wang↗

arXiv:2604.18701v3 Announce Type: replace-cross Abstract: Local prediction-error-based curiosity rewards focus on the current transition without considering the world model's cumulative prediction error across all visited transitions. We introduce Curiosity-Critic, which grounds its intrinsic reward in the improvement of this cumulative objective, and show that it admits a tractable per-step surrogate: the difference between the current prediction error and the asymptotic error baseline of the current state transition. We estimate this error baseline online with a learned critic co-trained alongside the world model; since the critic only has to learn how hard a transition is to predict, its estimate of the irreducible noise floor converges well before the world model saturates, redirecting exploration toward learnable transitions. The reward is higher for learnable transitions and collapses toward zero for stochastic ones, thereby separating epistemic (reducible) from aleatoric (irreducible) prediction error online. Prior prediction-error curiosity formulations, from Schmidhuber (1991) to learned-feature-space variants, emerge as special cases corresponding to specific approximations of this error baseline. Experiments on a stochastic grid world show that Curiosity-Critic outperforms prediction-error, visitation-count, and Random Network Distillation methods in training speed and final world model accuracy.

阅读与讨论 → 访问原文 →

21.

arXiv (CS.CL) 2026-06-19 DOI: arXiv:2606.19501

DeXposure-Claw: An Agentic System for DeFi Risk Supervision

作者:

Aijie Shu↗Bowei Chen↗Wenbin Wu↗Cathy Yi-Hsuan Chen↗Fengxiang He↗

Decentralized finance exposes supervisors to fast-moving, networked credit risks. General-purpose LLM agents fit this setting poorly: they over-read weak evidence and recommend high-stakes interventions, while existing evaluations offer no regulator-aligned way to measure the resulting false alarms. We introduce DeXposure-Claw, a forecast-grounded agentic supervision system that routes LLM decisions through structured evidence: (1) DeXposure-FM, a graph time-series foundation model, forecasts future exposure networks; (2) deterministic monitors and stress scenarios then turn those forecasts into typed alerts, attribution signals, and scenario evidence; and (3) data-health and confidence gates constrain escalation before DeXposure-Claw emits auditable supervisory tickets with rationales. We further develop DeXposure-Bench, a six-axis evaluation harness, whose decision axis scores tickets against a regulator-aligned absolute-loss ground truth and an explicit false-intervention rate. Experiments on five years of weekly real data fully support our system. Code is at https://github.com/EVIEHub/DeXposure-Claw.

阅读与讨论 → 访问原文 →

22.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.11709

RLCSD: Reinforcement Learning with Contrastive On-Policy Self-Distillation

作者:

Leyi Pan↗Shuchang Tao↗Yunpeng Zhai↗Lingzhe Zhang↗Zhaoyang Liu↗Bolin Ding↗Aiwei Liu↗Lijie Wen↗

On-policy self-distillation (OPSD) provides dense, token-level supervision for reasoning models by aligning a model's own distribution with the distribution it produces under privileged context, typically a verified solution. However, we show that the learning signal drawn from this distributional gap concentrates on style tokens rather than task-bearing ones, as the hinted model tends to produce more direct, shorter outputs. We term this pathology privilege-induced style drift, which destabilizes training or causes response length to shrink. To address this, we propose RLCSD (Reinforcement Learning with Contrastive on-policy Self-Distillation), which mitigates this drift by contrasting the teacher-student gap under a correct hint against that under a wrong hint, suppressing the style shift that conditioning on a hint tends to induce regardless of correctness, and yielding a signal that is more concentrated on task-bearing tokens. Experiments on Qwen3 (1.7B/4B/8B) and Olmo-3-7B-Think across mathematical and logical reasoning show that RLCSD consistently outperforms GRPO and prior OPSD methods. We further show that the contrastive principle is general: it plugs into existing OPSD methods to improve them, and its underlying insight extends to the broader cross-model on-policy distillation setting.

阅读与讨论 → 访问原文 →

23.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2606.19812

Human-on-the-Loop Orchestration for AI-Assisted Legal Discovery

作者:

Anushree Sinha↗Srivaths Ranganathan↗Abhishek Dharmaratnakar↗Debanshu Das↗

arXiv:2606.19812v1 Announce Type: new Abstract: Autonomous Large Language Model (LLM) agents are increasingly deployed in electronic discovery (e-discovery), where compounding errors across multi-step reasoning chains can constitute legal malpractice. Unlike single-turn retrieval, agentic workflows operating over privileged document corpora exhibit a class of failure we term "trajectory collapse": an early misclassification silently propagates, rendering an entire privilege review invalid. This paper makes three contributions. First, we propose a structured taxonomy of agentic failures in legal information retrieval, organized by functional stage. Second, we introduce a four-layer verification architecture – spanning planning, reasoning, execution, and uncertainty quantification – designed to intercept these failures before they compound. Third, we present a preliminary simulation study on a synthetic e-discovery corpus that demonstrates how mandatory Human-on-the-Loop (HOTL) escalation thresholds reduce privilege-waiver risk relative to fully autonomous baselines. Our results suggest that calibrated uncertainty thresholds can reduce privilege-waiver risk by up to 61% versus fully autonomous deployment, while routing fewer than one quarter of documents to attorney review.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.LG) 2026-06-11 DOI: arXiv:2606.11382

GLACIER: A Multimodal Student-Teacher Foundation Model for Molecular Property Prediction

作者:

Emily Nguyen↗Yongchan Hong↗Harsh Toshniwal↗Yan Liu↗Andreas Luttens↗

arXiv:2606.11382v1 Announce Type: new Abstract: Deep learning models facilitate the discovery of molecules with tailored properties among billions of candidate compounds. However, the computational burden to develop and deploy state-of-the-art models continuously increases, limiting their scalability. Most large-scale models are unimodal in nature and overlook the potential to leverage complementary molecular data modalities. To address these shortcomings, this paper introduces the Graph-Language Alignment for Chemical Inference and Exploration using Representations (GLACIER) model, a student-teacher framework that integrates molecular graphs, SMILES strings, and physicochemical descriptors to learn rich molecular embeddings. Our framework consists of three stages: (1) we pretrain three student encoders on 100,000 drug-like molecules: a message-passing neural network for molecular graphs, a transformer-based encoder for SMILES strings, and a multilayer perceptron for physicochemical descriptors, (2) we fuse these student modalities using a novel Finsler geometry-aware module, and (3) distill complementary knowledge from large teacher models, including MiniMol and MolFormer, into a single lightweight model via contrastive learning. We demonstrate that GLACIER is a robust framework that delivers high predictive performance and computational efficiency in complex molecular property prediction tasks. Our code is publicly available at https://github.com/eemokey/glacier.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.AI) 2026-06-18 DOI: arXiv:2606.18303

A Link between Shock-wave Theory and Symmetry-reduced Stochastic Gradient Descent for Artificial Neural Networks

作者:

Taiki Miyagawa↗

arXiv:2606.18303v1 Announce Type: cross Abstract: We develop a mathematically explicit link between shock-wave theory and the symmetry-quotiented learning dynamics of stochastic gradient descent, drawing on differential geometry, Lie group theory, and fluid mechanics. Specifically, after quotienting parameter symmetries and applying local-entropy coarse-graining, the effective dynamics satisfy a viscous Hamilton–Jacobi equation on the quotient manifold. Moreover, under the assumption that the raw parameter dynamics can be summarized by a gradient field on the quotiented space, the gradient of the coarse-grained loss function obeys a Burgers-type equation, and shock formation can be established rigorously. We apply our theory to multilayer perceptrons, convolutional neural networks, Transformers, and mean-field networks, and show that they obey the Hamilton–Jacobi or Burgers-type equations. We conjecture that this framework also yields practical diagnostics for deep learning. In architectures such as Transformers, raw parameter norms are often distorted by symmetry redundancy and may therefore be misleading, whereas symmetry-corrected quotient observables provide a principled basis for monitoring, forecasting, and controlling training-phase transitions.

阅读与讨论 → 访问原文 →