探索全球前沿学术脉络

01.

arXiv (CS.AI) 2026-06-12 DOI: arXiv:2606.13197

ARMOR-MAD: Adaptive Routing for Heterogeneous Multi-Agent Debate in Large Language Model Reasoning

作者:

Fuqiang Niu↗Bowen Zhang↗

arXiv:2606.13197v1 Announce Type: new Abstract: Multi-agent debate (MAD) can improve large language model reasoning, but fixed debate pipelines often waste computation and can amplify correlated errors among similar agents. We propose ARMOR-MAD, a training-free heterogeneous MAD framework that treats debate as conditional computation. ARMOR-MAD combines three components: Pre-debate Agreement Routing (PAR) decides whether independently generated Round-0 answers require debate; Early Agreement Stopping Evaluator (EASE) stops debate after convergence; and Semantic Outlier Detection (SOD) down-weights abnormal final answers during aggregation. Across MATH Level 5, GSM8K, MMLU, and MMLU-Pro, ARMOR-MAD consistently improves over fixed-round heterogeneous debate with the same model pool, reaching 65.5\%, 96.5\%, 90.0\%, and 81.5\% accuracy, respectively. The results suggest that genuine model heterogeneity and agreement-based control are both important for making MAD more accurate and efficient.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.CV) 2026-06-17 DOI: arXiv:2606.17256

Contrastive Action-Image Pre-training for Visuomotor Control

作者:

Yuvan Sharma↗Dantong Niu↗Anirudh Pai↗Zekai Wang↗Zhuoyang Liu↗Baifeng Shi↗Stefano Saravalle↗Boning Shao↗Ruijie Zheng↗Jing Wang↗Konstantinos Kallidromitis↗Yusuke Kato↗…

Existing vision encoders for robotics face a fundamental bottleneck: robotic datasets lack the scale necessary for large-scale pre-training. Prior work circumvents this data scarcity by turning to internet-scale image and language data or egocentric human video. While these models show promise, neither paradigm learns from paired vision and action data, which downstream visuomotor control policies require. However, robot trajectories, the most direct source of this paired signal, are not available at pre-training scale, motivating us to extract action signals from abundant human video instead. To this end, we introduce CAIP (Contrastive Action-Image Pre-training), a vision encoder that treats human hand poses from large-scale egocentric video as a proxy for end-effector actions. By extracting 3D hand keypoints, a representation that aligns naturally with downstream robot action spaces, CAIP learns a unified action-image representation through a contrastive objective. Leveraging 32,041 hours of egocentric human video and only 88 hours of robotic manipulation data, CAIP outperforms state-of-the-art vision encoders including DINOv2, SigLIP, MVP, and R3M. Evaluated on a challenging real-world dexterous manipulation setup using Dexmate Vega and Sharpa Wave hands, CAIP yields performance gains of more than 30% on tasks involving folding, pouring, and fine-grained manipulation. Our results show that our method of contrastive action-centric pre-training yields a scalable path to achieving robust visual representations better suited for physical interaction.

阅读与讨论 → 访问原文 →

03.

medRxiv (Medicine) 2026-06-22 DOI: HASH:b3d6abddfc06053c524bb4c6abb61d29

COVID-19 containment policies and hyperglycemia in pregnancy: correlation with the Stringency Index in a nationwide Belgian cohort

作者:

Costa↗Vercoutere↗Alexander↗Boulvain↗Derisbourg↗Lamy↗Delbaere↗Racape↗

Background During the COVID-19 pandemic, gestational diabetes (GD) prevalence showed variable changes across regions, with most reporting increases and others decreases; however, its association with perinatal outcomes in Belgium remains unknown. We aimed to compare the prevalence of hyperglycemia in pregnancy (HIP) in 2020 versus 2019 and examined the correlation between HIP prevalence and pandemic-related restrictions measured by the Stringency Index (SI) and evaluate neonatal weight percentiles changes. Methods: We included all singleton live births in Belgium in 2019 and 2020 from Belgian birth registry data. We compared monthly proportions of HIP prevalence and Small for gestational age (SGA) and Large for gestional age (LGA) newborns in 2019 and 2020. Crude and adjusted odds ratios (ORs, aORs) were estimated with logistic and multinomial regression. The Spearman correlation coefficient was used to assess the correlation between the monthly average SI and the monthly aORs of HIP. Results: For deliveries from January to June 2020, no significant differences in HIP prevalence were observed compared with 2019. From July to December 2020, there was a significant increase in HIP, with peaks in July (GD screening in April) (aOR 1.41, 1.26-1.58) and November (GD screening in August) (aOR 1.33, 95% CI 1.18-1.49). There was no significant change in neonatal weight percentiles. The Spearman correlation coefficient between the SI and HIP aORs was 0.86 (p = 0.02). Conclusion During the pandemic, we observed an increase in the prevalence of HIP, compared to 2019, without a measurable impact on LGA or SGA newborns. The aOR of HIP in a given month was strongly correlated with the corresponding SI.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.AI) 2026-06-15 DOI: arXiv:2510.02695

RAMAC: Multimodal Risk-Aware Offline Reinforcement Learning and the Role of Behavior Regularization

作者:

Kai Fukazawa↗Kunal Mundada↗Iman Soltani↗

arXiv:2510.02695v3 Announce Type: replace-cross Abstract: In safety-critical domains where online data collection is infeasible, offline reinforcement learning (RL) is attractive only if policies achieve high returns without catastrophic lower-tail risk. Prior work on risk-averse offline RL achieves safety at the cost of either (i) value/model-based pessimism or (ii) restricted policy classes that limit expressiveness, whereas diffusion/flow-based expressive generative policies have largely been used in risk-neutral settings. We introduce Risk-Aware Multimodal Actor-Critic (RAMAC), a simple, modular, model-free framework that couples an expressive generative actor (e.g., diffusion/flow) with a distributional critic and optimizes a composite objective that combines Conditional Value-at-Risk (CVaR) with behavioral cloning (BC), enabling risk-sensitive learning in complex multimodal scenarios. Since out-of-distribution (OOD) actions are a major driver of catastrophic failures in offline RL, we further provide an objective-level analysis showing that controlling behavior divergence via BC suppresses OOD actions and stabilizes CVaR. Instantiating RAMAC with a diffusion actor, we illustrate these insights on a 2-D risky bandit and evaluate on Stochastic-D4RL, observing consistent gains in $\mathrm{CVaR}_{0.1}$ while maintaining strong returns. The code and experimental results are available on the \href{https://kaifukazawa.github.io/ramac-project/} {project website}

阅读与讨论 → 访问原文 →

05.

medRxiv (Medicine) 2026-06-22 DOI: HASH:8b3b0ac0c6d68b53b70c81ef0a11ece2

GCH1 p.Ser80Asn Confers Risk for Parkinson's Disease in East Asian Populations

作者:

Tay↗Y. W↗Lee↗A. L↗Schee↗J. P↗Lin↗C. H↗Tan↗E. K↗Shin↗J. H↗…

Introduction: GCH1 has been implicated in Parkinson's disease (PD), but its risks variants and associations are not well defined. Objectives: To investigate the clinical relevance and PD risk associated with the GCH1 p.Ser80Asn variant. Methods: We first identified a segregating GCH1 p.Ser80Asn variant in a Malaysian Chinese PD family via whole genome sequencing (WGS). We assessed its risk association using multi-ancestry WGS data from the Global Parkinson's Genetics Program (GP2) (n=22,372PD vs n=8,826Controls) and meta-analysis of East Asian (EAS) cohorts (n=4,712PD vs 38,733Controls). Clinico-demographic details of affected variant carriers were collated. Results: The GCH1 p.Ser80Asn variant was enriched in GP2 EAS PD populations (n=9/2,757; 0.33%) but not detected in other ancestries. Meta-analysis revealed increased PD risk in EAS populations (odds ratio:5.1; 95%CI:2.3-10.7; p=2.89x10-5). Affected carriers (mean age at onset:56.3+-12.5 years) had additional occurrence of dystonia, while dementia was rare. Conclusions: The GCH1 p.Ser80Asn variant is a rare, EAS-enriched risk variant for PD.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.AI) 2026-06-17 DOI: arXiv:2512.25065

Vulcan: Instance-specialized, Verifiable Systems Heuristics Through LLM-driven Search

作者:

Rohit Dwivedula↗Divyanshu Saxena↗Sujay Yadalam↗Eric Hayden Campbell↗Daehyeok Kim↗Aditya Akella↗

arXiv:2512.25065v2 Announce Type: replace-cross Abstract: Systems resource management tasks rely primarily on hand-designed heuristics. However, growing hardware heterogeneity and workload diversity require heuristics specialized to particular deployment instances, making manual design expensive and difficult to scale. In this paper, we explore how to synthesize systems heuristics using LLMs. The main challenge is ensuring that generated heuristics execute safely, integrate correctly with the surrounding system, and still achieve strong performance. We propose Vulcan, a framework that identifies LLM-friendly interfaces that isolate core decision logic from the rest of the implementation. With Vulcan, LLM-generated code is restricted to simple stateless decision functions, while trusted runtime abstractions provide rich derived statistics for meaningful policy exploration without system-integration bugs. To ensure execution safety, LLMs synthesize heuristics in a restricted language, Anvil, that guarantees important properties by construction. We evaluate Vulcan across three well-studied domains and demonstrate up to 4.9x higher savings for spot-VM scheduling, up to 2x lower miss ratios for cache eviction, and up to 10% higher application performance for tiered-memory systems, while ensuring execution safety throughout.

阅读与讨论 → 访问原文 →

07.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.16062

Auditing Reward Hackability in Code RL Training Environments

作者:

Shreshth Rajan↗

arXiv:2606.16062v1 Announce Type: new Abstract: We measure the rate at which code RL environments accept incorrect solutions as correct. On a 49-task sample of SWE-bench Verified, 28.5% of tasks have test suites weak enough that a Docker-verified incorrect patch passes them. On 20 R2E-Gym tasks across 6 repositories, the same pipeline at single-shot exploit generation yields 25.0%. A random-effects meta-analysis over 134 frontier model submissions to SWE-bench Verified finds, within the same human-rated difficulty stratum, model Pass@1 is +14.14 percentage points higher on flagged-hackable tasks than on robust ones (95% CI [+11.80, +16.48]; one-sided p < 10^-6; I^2 = 0%; 123 of 134 models positive). We then describe a procedure for hardening the broken tasks. An inline LLM judge with a Docker gold-sanity gate runs each generated test against the gold solution before the judge is consulted. On the 11 broken tasks in the audit, the gate flags 65 of 105 decisive LLM-generated tests as failing on the gold patch itself, a 61.9% per-augmentation defect rate the LLM judge alone misses. With diversity-biased retry, the loop converges 9 of 11 tasks to a gated upgrade.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.LG) 2026-06-15 DOI: arXiv:2509.24710

MAD: Manifold Attracted Diffusion

作者:

Dennis Elbr\"achter↗Giovanni S. Alberti↗Matteo Santacesaria↗

arXiv:2509.24710v3 Announce Type: replace-cross Abstract: Score-based diffusion models are a highly effective method for generating samples from a distribution of images. We consider scenarios where the training data comes from a noisy version of the target distribution, and present an efficiently implementable modification of the inference procedure to generate noiseless samples. Our approach is motivated by the manifold hypothesis, according to which meaningful data is concentrated around some low-dimensional manifold of a high-dimensional ambient space. The central idea is that noise manifests as low magnitude variation in off-manifold directions in contrast to the relevant variation of the desired distribution which is mostly confined to on-manifold directions. We introduce the notion of an extended score and show that, in a simplified setting, it can be used to reduce small variations to zero, while leaving large variations mostly unchanged. We describe how its approximation can be computed efficiently from an approximation to the standard score and demonstrate its efficacy on toy problems, synthetic data, and real data.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.18867

Strategic Feature Selection

作者:

Jivat Neet Kaur↗Pratik Patil↗Divya Shanmugam↗Emma Pierson↗Michael I. Jordan↗Nika Haghtalab↗Meena Jagadeesan↗Ahmed Alaa↗Serena Wang↗

arXiv:2606.18867v1 Announce Type: new Abstract: When algorithmic predictors inform resource allocation in high-stakes domains such as healthcare, these predictors must account for strategic manipulation of input features. The typical solution is to redesign the predictor itself to explicitly account for strategic interactions. In practice, however, decision makers are often constrained to adjusting coarser levers within existing prediction pipelines. For example, healthcare organizations often select which features to exclude based on perceived manipulability, while using standard regularization procedures to shrink the coefficients of retained features. In this work, we initiate a formal study of strategic classification through feature selection and its interaction with ridge regularization. Our main finding is that excluding individual features based on their manipulability alone is generally suboptimal. We provide a fine-grained characterization of the performance of a feature subset under optimal regularization, yielding new insights for policy design. Motivated by this characterization, we develop a practical algorithm for jointly choosing the feature set and the level of ridge regularization. Through a real-world case study on a healthcare payments benchmark, we illustrate how our algorithm can guide the design of coarse policy levers in practice. Our results provide a principled, practical framework for mitigating the effects of strategic behavior in algorithmic decision-making systems.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.LG) 2026-06-17 DOI: arXiv:2603.02159

Instrumental and Proximal Causal Inference with Gaussian Processes

作者:

Yuqi Zhang↗Krikamol Muandet↗Dino Sejdinovic↗Edwin Fong↗Siu Lun Chau↗

arXiv:2603.02159v2 Announce Type: replace-cross Abstract: Instrumental variable (IV) and proximal causal learning (Proxy) methods are central frameworks for causal inference in the presence of unobserved confounding. Despite substantial methodological advances, existing approaches rarely provide reliable epistemic uncertainty (EU) quantification. We address this gap through a Deconditional Gaussian Process (DGP) framework for uncertainty-aware causal learning. Our formulation recovers popular kernel estimators as the posterior mean, ensuring predictive precision, while the posterior variance yields principled and well-calibrated EU. Moreover, the probabilistic structure enables systematic model selection via marginal log-likelihood optimization. Empirical results demonstrate strong predictive performance alongside informative EU quantification, evaluated via empirical coverage frequencies and decision-aware accuracy rejection curves. Together, our approach provides a unified, practical solution for causal inference under unobserved confounding with reliable uncertainty.

阅读与讨论 → 访问原文 →

11.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2505.16988

MASLab: A Unified and Comprehensive Codebase for LLM-based Multi-Agent Systems

作者:

Rui Ye↗Keduan Huang↗Qimin Wu↗Yuzhu Cai↗Tian Jin↗Xianghe Pang↗Xiangrui Liu↗Jiaqi Su↗Chen Qian↗Bohan Tang↗Kaiqu Liang↗Jiaao Chen↗…

LLM-based multi-agent systems (MAS) have demonstrated significant potential in enhancing single LLMs to address complex and diverse tasks in practical applications. Despite considerable advancements, the field lacks a unified codebase that consolidates existing methods, resulting in redundant re-implementation efforts, unfair comparisons, and high entry barriers for researchers. To address these challenges, we introduce MASLab, a unified, comprehensive, and research-friendly codebase for LLM-based MAS. (1) MASLab integrates over 20 established methods across multiple domains, each rigorously validated by comparing step-by-step outputs with its official implementation. (2) MASLab provides a unified environment with various benchmarks for fair comparisons among methods, ensuring consistent inputs and standardized evaluation protocols. (3) MASLab implements methods within a shared streamlined structure, lowering the barriers for understanding and extension. Building on MASLab, we conduct extensive experiments covering 10+ benchmarks and 8 models, offering researchers a clear and comprehensive view of the current landscape of MAS methods. MASLab will continue to evolve, tracking the latest developments in the field, and invite contributions from the broader open-source community.

阅读与讨论 → 访问原文 →

12.

arXiv (CS.AI) 2026-06-12 DOI: arXiv:2606.13079

The Emergence of Autonomous Penetration Capabilities in Large Language Model-Powered AI Systems

作者:

Jiaqi Luo↗Jiarun Dai↗Zhile Chen↗Jia Xu↗Weibing Wang↗Yawen Duan↗Brian Tse↗Geng Hong↗Xudong Pan↗Yuan Zhang↗Min Yang↗

arXiv:2606.13079v1 Announce Type: cross Abstract: Nowadays, the autonomous execution of cyberattacks capable of causing substantial real-world harm is widely regarded as one of the critical red lines that frontier AI systems must not cross. Within this broader red-line scenario, autonomous penetration represents a core enabling capability and subtask: the ability of LLM-powered AI systems to independently conduct adversarial operations against a target server without human intervention, identify and exploit vulnerabilities, and obtain unauthorized access or control. A growing body of work has sought to assess the autonomous penetration capabilities of AI systems. However, existing evaluations often employ opaque methodologies, rely on unrealistic or overly simplified penetration-testing scenarios, or provide LLMs with excessive prior knowledge and task-specific guidance, and cannot accurately capture the extent to which modern AI systems can autonomously perform this core capability within broader high-impact cyberattack scenarios. To address these limitations, we construct a new autonomous penetration evaluation framework consisting of two components: target servers and agent scaffolding. Specifically, on the target-server side, we design two levels of target environments based on the number of secure services without known vulnerabilities deployed alongside a vulnerable service: Tier~1 (one secure service) and Tier~2 (three secure services), resulting in a total of 300 target servers. Meanwhile, the agent scaffolding adopts a general-purpose agent architecture equipped with a set of general-purpose cybersecurity tools, without any target-specific prior knowledge. We evaluate 19 open-weight and proprietary LLMs, and find that current models achieve penetration success rates ranging from 10.7% to 69.3%. Moreover, we observe that autonomous penetration capability continues to improve alongside advances in overall model capability.

阅读与讨论 → 访问原文 →

13.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.15273

Feature Attribution in Directed Acyclic Graphs Using Edge Intervention

作者:

Qiheng Sun↗Junxu Liu↗Xiaokai Mao↗Haocheng Xia↗Jinfei Liu↗Kui Ren↗Haibo Hu↗

arXiv:2606.15273v1 Announce Type: new Abstract: Shapley value-based feature attribution methods face challenges in scenarios involving complex feature interactions and causal relationships, even when a causal structure is provided. Existing methods typically adopt a node-centric view, attributing importance solely to individual features. Consequently, they often fail to simultaneously capture the externality and exogenous influence of features, leading to unreasonable interpretations. To overcome these limitations, we propose a novel feature attribution method called DAG-SHAP, which is based on edge intervention. DAG-SHAP treats each feature edge as an individual attribution object, ensuring that both externality and exogenous contributions of features are appropriately captured. Additionally, we introduce an approximation method for efficiently computing DAG-SHAP. Extensive experiments on both real and synthetic datasets validate the effectiveness of DAG-SHAP. Our code is available at https://github.com/ZJU-DIVER/DAG-SHAP.

阅读与讨论 → 访问原文 →

14.

arXiv (CS.LG) 2026-06-15 DOI: arXiv:2606.14668

When to Write and When to Suppress: Route-Specialized Dual Adapters for Memory-Assisted Knowledge Editing

作者:

Yining Huang↗

arXiv:2606.14668v1 Announce Type: new Abstract: Knowledge editing systems must update selected facts while preserving nearby but irrelevant behavior. This paper studies this problem in a memory-assisted setting where an edit memory is retrieved at inference time and a parameter-efficient adapter corrects the model's object preference. We argue that the central design question is not only how to write an edit, but also when to suppress it. We introduce \method{}, a route-specialized dual-adapter editor. A relevance router first decides whether a prompt should receive an edit memory. Routed prompts use an edit adapter trained to prefer the new object over the original object; unrouted non-direct prompts use a separate locality adapter trained to preserve or restore the original-object preference. We evaluate \method{} on three 1,000-case protocols, \cf{}, \zsre{}, and \mquake{}, under the same memory protocol and two 7B/8B base models. On Llama-3.1-8B-Instruct, \method{} obtains the best overall probability-preference accuracy on all three benchmarks: 0.8180 on \cf{}, 0.8946 on \zsre{}, and 0.9922 on \mquake{}. The same trend holds on Qwen3-8B. Router ablations show that the relevant memory boundary differs across datasets: a lexical neural router is safest on \cf{}, while BGE embedding routing is better on \zsre{} and \mquake{}. Component and module ablations show that the gain mainly comes from separating edit injection from off-route suppression rather than from simply increasing LoRA capacity.

阅读与讨论 → 访问原文 →

15.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2605.27023

Boosting Knowledge Graph Foundation Models via Enhanced Negative Sampling

作者:

Yinan Liu↗Wenjin Xu↗Zhiyuan Zha↗Xiaochun Yang↗Bin Wang↗

arXiv:2605.27023v2 Announce Type: replace Abstract: Knowledge graphs (KGs) have become the core backbone of numerous downstream tasks such as question answering and recommender systems. However, despite all this, KGs are often very incomplete. To perform zero-shot knowledge graph completion in unseen KGs, which have different relational vocabularies from those used for pre-training, KG foundation models (KGFMs) receive a wide range of attention. Existing KGFMs often perform training using random negative triples, which are constructed by replacing the head or tail entity of a positive triple with a random entity. However, these negative triples are often constructed with limited quality, providing weak supervision for KGFM training. In this paper, we propose a simple yet effective adaptive negative sampling approach, KMAS, to enhance existing KGFMs. KMAS constructs hard negative triples through the updated relation embeddings generated from the existing KGFM's relation encoder. To further adaptively align with the evolving capability of the KGFM during the training process, KMAS adjusts the ratio of hard negative triples dynamically throughout the whole training process: after a warmup phrase, it increases the ratio linearly and then decreases linearly. Extensive experiments are conducted over 44 data sets. Experimental results demonstrate that our proposed negative sampling method can enhance many SOTA KGFMs without requiring excessive additional time or memory consumption.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.19101

Structure Over Nonlinearity: Explicit Interaction Architectures for Dynamical Learning

作者:

Augusto Sarti↗

arXiv:2606.19101v1 Announce Type: cross Abstract: Most learning architectures for dynamical systems rely on generic nonlinear function approximation, often requiring high model complexity to capture structured behaviors. In this work, we propose an alternative paradigm in which modeling capability arises primarily from structure rather than from expressive nonlinearities. We introduce a class of explicit structured dynamical units based on wave-inspired interaction structures with internal state. Inspired by wave-based computational principles, the proposed units adopt a strictly causal organization that eliminates algebraic loops, yielding fully explicit models that can be evaluated without implicit solvers. Stacking such units produces layered dynamical architectures with emergent hierarchical behavior. Through experiments on a nonlinear system identification task, we show that depth improves both representation quality and generalization, even under limited parameter optimization. In particular, the proposed architectures produce informative internal representations even under readout-only fitting, indicating that useful dynamical structure emerges from the organization of interactions prior to substantial parameter optimization. These results suggest that structure-first design provides a viable and effective alternative to conventional black-box approaches for learning dynamical systems, highlighting the role of interaction structure as a primary source of model expressivity.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.CL) 2026-06-19 DOI: arXiv:2606.19946

GEMS: Geometric Constraints Enable Multi-Semantic Superposition in LLMs

作者:

Yu Deng↗

Activation steering controls model behavior by modifying intermediate hidden states at inference time without retraining. Existing methods handle only single-direction injection; when multiple semantic directions are superposed without constraints, the model collapses. We show that this collapse decomposes into two independently acting sources: distributional deviation, where additive perturbations accumulate in norm across layers and drive activations outside the training distribution, and directional interference, where non-orthogonal semantic vectors mutually dampen when superposed. These two sources define the design constraints that any training-free multi-directional intervention must address. As one instantiation of these principles, we propose GEMS, a training-free method that maps each source to a corresponding geometric constraint: norm-preserving weighted superposition and targeted attention-pathway injection for distributional deviation, and real-time orthogonalization for directional interference. On GSM8K, injecting three concurrent non-mathematical directions preserves accuracy at 98% (baseline 92%), while unconstrained addition collapses to 4%; on Wikitext-2, the same injection incurs only 2.2% PPL increase. Component ablation isolates the causal role of each constraint, and layer-level probes confirm that orthogonalized signals survive the FFN pathway and reach the output distribution with semantic specificity. Qualitative steering effects transfer across architectures from 3B to 31B.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2503.07459

MedicalAgentsBench for Complex Medical Reasoning: Comparing Internalized Reasoning Models versus Externalized Agent-based Frameworks

作者:

Yanjun Shao↗Xiangru Tang↗Jiwoong Sohn↗Jiapeng Chen↗Yuxuan Liao↗Jiayi Zhang↗Jinyu Xiang↗Fang Wu↗Yilun Zhao↗Chenglin Wu↗Wenqi Shi↗Arman Cohan↗…

Complex medical reasoning requires integrating heterogeneous clinical evidence across multiple inference steps. Large language models (LLMs) now approach this through two routes: internalized reasoning and externalized agent scaffolding (frameworks that decompose problems collaboratively amongst multiple LLMs). To determine whether these routes are exclusive or complementary, we introduce MedicalAgentsBench, a filtered benchmark of 862 complex clinical questions drawn from the union of eight medical datasets via difficulty-aware curation and contamination screening. Evaluating three internalized reasoning models (DeepSeek-R1, o1-mini, and o3-mini), seven base models, and nine externalized agent-based methods, we find that internalized and externalized approaches each independently improve performance, and that their benefits compound: the highest accuracy is achieved by layering agent workflows onto an internalized reasoning model (i.e., o3-mini + MDAgents with 35.1%). Pareto analysis shows this combination dominates the cost-performance frontier; moreover, lightweight optimization on inexpensive models offers an entry point for resource-constrained settings. Our benchmark is at https://github.com/gersteinlab/MedicalAgentsBench.

阅读与讨论 → 访问原文 →

19.

bioRxiv (Bioinfo) 2026-06-20 DOI: HASH:acdc961f91405ee23cdf99387b8794d4

The recount3 Python package for programmatic access to uniformly processed RNA-seq data

作者:

Alsalihi↗Flight↗R. M↗Moseley↗H. N. B↗

The recount3 online resource provides tens of thousands of uniformly processed RNA-seq samples across human and mouse from major sequencing repositories like the Sequence Read Archive. While access to these datasets has traditionally been centered in the R/Bioconductor ecosystem, the growing prominence of Python in bioinformatics and machine learning necessitates native, efficient tooling for Python users. Therefore, we present the recount3 Python package with robust application programming interface (API) and command-line interface (CLI) for discovering, downloading, and materializing recount3 resources. The software orchestrates uniform resource locator (URL) resolution, persistent on-disk caching, and the automatic parsing of data into analysis-ready data structures, including Pandas DataFrames and BiocPy RangedSummarizedExperiment objects. The recount3 Python package drastically lowers the barrier to entry for large-scale utilization of RNA-seq data in Python-based computational pipelines, bridging the gap between massive public transcriptomic data and modern machine learning ecosystems.

阅读与讨论 → 访问原文 →

20.

medRxiv (Medicine) 2026-06-11 DOI: HASH:c1a57498dc29b202287a19079ac0d6c4

Validity and Limitations of the Empatica E4 Wristband for Autonomic and Thermoregulatory Sleep Monitoring Against Concurrent Polysomnography: A Wearanize+ Dataset Study

作者:

Parry↗Y. D↗Briganti↗

The Empatica E4 wristband provides continuous multi-modal physiological monitoring including blood volume pulse (BVP), electrodermal activity (EDA) and skin temperature (TEMP) but its validity for sleep-stage-specific autonomic and thermoregulatory monitoring has not been systematically evaluated against concurrent polysomnography (PSG). Using the Wearanize+ dataset which provides synchronised PSG, Empatica E4, and Zmax EEG recordings from 100 home-recorded participants; a systematic validation of Empatica E4 physiological signals against PSG ground truth across five sleep stages was conducted. Of 100 participants, 92 had Empatica data; 69 met Zmax EEG signal quality criteria and formed the analysis sample. Heart rate (HR) from the pre-computed Empatica HR channel showed valid stage-specific patterns (Wake: 70.9 bpm, N3: 61.2 bpm) and moderate inter-device MeanNN correspondence with PSG ECG (Spearman r=0.35-0.42 across stages). Skin temperature showed the expected thermoregulatory pattern (Wake: 33.92C, N3: 35.48C) and is recommended for downstream analyses. Tonic EDA showed an inverted stage pattern attributable to wrist sweat accumulation during deep sleep, representing a known confound for wrist-worn EDA during sleep. Phasic EDA showed plausible patterns and may be used with caution. These findings establish a validated feature set for Empatica E4 sleep research and directly inform multimodal psychiatric biomarker studies using the Wearanize+ dataset.

阅读与讨论 → 访问原文 →

21.

arXiv (math.PR) 2026-06-11 DOI: arXiv:2606.11271

The $K$-th nearest neighbor random walk on a Poisson point process gets trapped

作者:

Anne-Laure Basdevant↗David Coupier↗Jean-Baptiste Gou\'er\'e↗Marie Th\'eret↗

arXiv:2606.11271v1 Announce Type: new Abstract: The $K$-th nearest neighbor random walk $(X_n)_{n \geq 0}$ on a homogeneous Poisson point process $\chi$ on $\R^d$ ($d\geq 1$), starts at the origin and at each step picks its next Poisson point among its closest neighbors according to i.i.d. labels having the same distribution as $K$. Our main result (Theorem 1) states that the number of Poisson points visited by $(X_n)_{n \geq 0}$ admits an exponential decay whenever the random variable $K$ has a bounded support (BS). In particular, the $K$-th nearest neighbor random walk visits finitely many Poisson points if and only if $K$ satisfies Assumption (BS). To prove it, we introduce the key notion of pioneer point which allows us to deal with the region of $\R^d$ already explored by $(X_n)_{n \geq 0}$. Still under Assumption (BS), we also prove an exponential decay for the Euclidean length of the trajectory performed by $(X_n)_{n \geq 0}$ (Theorem 2). Finally, and quite surprisingly, we exhibit an example of label distribution with bounded support for which the $K$-th nearest neighbor random walk discovers new Poisson points after a number of steps whose tail distribution is at least polynomial (Theorem 3).

阅读与讨论 → 访问原文 →

22.

arXiv (CS.AI) 2026-06-11 DOI: arXiv:2604.13733

Vision-Language-Action Jump-Starting for Reinforcement Learning Robotic Agents

作者:

Angelo Moroncelli↗Roberto Zanetti↗Marco Maccarini↗Loris Roveda↗

arXiv:2604.13733v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) enables high-frequency, closed-loop control for robotic manipulation, but scaling to long-horizon tasks with sparse or imperfect rewards remains difficult due to inefficient exploration and poor credit assignment. Vision-Language-Action (VLA) models leverage large-scale multimodal pretraining to provide generalist, task-level reasoning, but current limitations hinder their direct use in fast and precise manipulation. In this paper, we propose Vision-Language-Action Jump-Starting (VLAJS), a method that bridges sparse VLA guidance with on-policy RL to improve exploration and learning efficiency. VLAJS treats VLAs as transient sources of high-level action suggestions that bias early exploration and improve credit assignment, while preserving the high-frequency, state-based control of RL. Our approach augments Proximal Policy Optimization (PPO) with a directional action-consistency regularization that softly aligns the RL agent's actions with VLA guidance during early training, without enforcing strict imitation, requiring demonstrations, or relying on continuous teacher queries. VLA guidance is applied sparsely and annealed over time, allowing the agent to adapt online and ultimately surpass the guiding policy. We evaluate VLAJS on six challenging manipulation tasks: lifting, pick-and-place, peg reorientation, peg insertion, poking, and pushing in simulation, and validate a subset on a real Franka Panda robot. VLAJS consistently outperforms PPO and distillation-style baselines in sample efficiency, reducing required environment interactions by over 50% in several tasks. Real-world experiments demonstrate zero-shot sim-to-real transfer and robust execution under clutter, object variation, and external perturbations.

阅读与讨论 → 访问原文 →

23.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.18844

Learning from Your Own Mistakes: Constructing Learnable Micro-Reflective Trajectories for Self-Distillation

作者:

Zhilin Huang↗Hang Gao↗Ziqiang Dong↗Yuan Chen↗Yifeng Luo↗Chujun Qin↗Jingyi Wang↗Yang Yang↗Guanjun Jiang↗

arXiv:2606.18844v1 Announce Type: new Abstract: Self-distillation improves reasoning in large language models by using the model's own rollouts as training signal, typically through implicit logit-level alignment that minimizes KL divergence toward a privileged target distribution. However, because this supervision is generated via uncontrolled sampling, it provides no diagnostic insight into the model's specific errors or corrective guidance for its individual failure patterns. Consequently, the model learns to imitate a privileged distribution rather than receiving fine-grained corrections that pinpoint where and why its reasoning fails. In this paper, we propose Trajectory-Augmented Policy Optimization (TAPO), which advances self-distillation from implicit distributional alignment to explicit trajectory construction. During RL training, the model produces both correct and incorrect rollouts to the same query, and TAPO leverages this contrastive structure to construct micro-reflective corrections, new training trajectories that retain the model's erroneous reasoning up to the point of failure, then insert a natural-language diagnosis and corrected reasoning guided by a correct reference from the same sampling group. Since each trajectory is anchored in the learner's own prefix and solutions, the corrective signal preserves the model's on-policy distribution to a greater extent than the position-wise alignment imposed by KL-based methods. To integrate these trajectories, TAPO introduces difficulty-aware candidate selection at the model's capability boundary and decoupled advantage estimation to prevent gradient contamination. Experiments on AIME 2024, AIME 2025, and HMMT 2025 show that TAPO achieves consistent improvements over GRPO under the same number of training steps. Further analysis demonstrates that TAPO strengthens both first-pass reasoning and error-correction effectiveness.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2603.23249

A Learning Method with Gap-Aware Generation for Heterogeneous DAG Scheduling

作者:

Ruisong Zhou↗Haijun Zou↗Li Zhou↗Chumin Sun↗Zaiwen Wen↗

arXiv:2603.23249v2 Announce Type: replace-cross Abstract: Efficient scheduling of directed acyclic graphs (DAGs) is a core problem in large-scale data-intensive computing systems, where query plans, data-processing workloads, and computation graphs consist of dependent tasks competing for limited heterogeneous resource pools. In practice, achieving high-performance execution requires schedulers to adapt across environments with varying resource pools and task types, while generating schedules under tight runtime budgets. We propose WeCAN, an end-to-end reinforcement learning framework for heterogeneous DAG scheduling that addresses task-pool compatibility coefficients and generation-induced optimality gaps. It adopts a two-stage single-pass design: a single forward pass produces task-pool scores and global parameters, followed by a generation map that constructs schedules without repeated network calls. Its weighted cross-attention encoder models task-pool interactions gated by compatibility coefficients, and is size-agnostic to environment fluctuations. Moreover, widely used list-scheduling maps can incur generation-induced optimality gaps from restricted reachability. We introduce an order-space analysis that characterizes the reachable set of generation maps via feasible schedule orders, explains the mechanism behind generation-induced gaps, and yields sufficient conditions for gap elimination. Guided by these conditions, we design a skip-extended realization with an analytically parameterized decreasing skip rule, which enlarges the reachable order set while preserving single-pass efficiency. Experiments on real-world TPC-H query DAGs, resource-intensive workload datasets, and ML-compiler computation graphs demonstrate improved makespan over strong baselines, with inference time comparable to classical heuristics and faster than multi-round neural schedulers.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.11722

ICA Lens: Interpreting Language Models Without Training Another Dictionary

作者:

Sida Liu↗Feijiang Han↗

Finding interpretable directions in language-model representations is critical for understanding and controlling model behavior. Sparse autoencoders (SAEs) have become the standard tool for this purpose, but using them as the default first lens often requires training, storing, and evaluating large overcomplete dictionaries. This bottleneck limits rapid exploration and raises a fundamental question: how much interpretable structure is already visible from activation geometry before training another neural dictionary? Our intuition is simple: many interpretable directions are selective on tokens, and these directions should look less Gaussian than random directions. We therefore revisit independent component analysis (ICA), a classical method for finding non-Gaussian directions, as a compact lens for language-model interpretability. We find that ICA has been underestimated for LLM interpretability, because prior uses often relied on off-the-shelf ICA implementations that are brittle on LLM activations and lacked systematic tools for inspecting and evaluating the recovered directions. To bridge these gaps, we introduce ICALens, the first practical workflow for stable, efficient, and auditable ICA analysis of LLM representations. It combines an optimized GPU-parallel FastICA pipeline with LLM-specific stability recipes and better fitting diagnostics, enabling efficient and reliable layer-wise analysis. Across GPT-2 Small, Gemma 2 2B, and Qwen 3.5 2B Base, ICALens efficiently recovers compact, human-interpretable directions without per-layer gradient-based dictionary training. On SAEBench, ICA is competitive with public SAEs in sparse probing and outperforms them in targeted probe perturbation under small-to-medium budgets. These results suggest that ICA should not be viewed as a weak baseline, but as an efficient and complementary first lens for exploring language-model representations.

阅读与讨论 → 访问原文 →