中 EN
登录 注册账户

Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达，利用大语言模型自动生成交叉领域文献分析简报。

登录 注册账户
01.
Nature Medicine 2026-06-12

General-purpose large language models outperform specialized clinical AI tools on medical benchmarks

作者:
Krithik Vishwanath↗

Specialized clinical artificial intelligence (AI) tools are entering medical practice despite scarce independent evaluation. We quantitatively evaluate two clinical AI tools, OpenEvidence and UpToDate Expert AI, built on large language models (LLMs) against three frontier LLMs: GPT-5.2, Gemini 3.1 Pro and Claude Opus 4.6. Our evaluation has three stages: (1) 500 MedQA questions testing medical knowledge, (2) 500 HealthBench items measuring alignment with clinicians and (3) the real clinical queries (RCQ) benchmark, built from 100 de-identified queries from physicians to a general-purpose language model in a live clinical environment. For the RCQ benchmark, 12 US clinicians performed randomized, blinded review of model outputs, producing 1,800 model–question annotations. Frontier LLMs outperformed clinical AI tools in all three evaluations. Clinical AI tools performed comparably to auto-enabled Google Search AI Overview on the RCQ. These findings highlight the need for independent, real-world evaluation of AI tools before they enter clinical settings. In an independent evaluation, frontier large language models outperformed specialized clinical artificial intelligence tools on medical knowledge, clinician alignment and real-world clinical queries.

阅读与讨论 → 访问原文 →
02.
arXiv (CS.CL) 2026-06-11

FOCUS: DLLMs Know How to Tame Their Compute Bound

作者:
Kaihua Liang↗Xin Tan↗An Zhong↗Hong Xu↗Marco Canini↗

Diffusion Large Language Models (DLLMs) offer a compelling alternative to Auto-Regressive models, but their deployment is constrained by high decoding cost. In this work, we identify a key inefficiency in DLLM decoding: while computation is parallelized over token blocks, only a small subset of tokens is decodable at each diffusion step, causing most compute to be wasted on non-decodable tokens. We further observe a strong correlation between attention-derived token importance and token-wise decoding probability. Based on this insight, we propose FOCUS, an inference system designed for DLLMs. By dynamically focusing computation on decodable tokens and evicting non-decodable ones on-the-fly, FOCUS increases the effective batch size, alleviating compute limitations and enabling scalable throughput. Empirical evaluations demonstrate that FOCUS achieves up to 3.52$\times$ throughput improvement over the production-grade engine LMDeploy in large-batch settings, while preserving or improving generation quality across multiple benchmarks.

阅读与讨论 → 访问原文 →
03.
arXiv (CS.AI) 2026-06-16

Parallel Test-Time Scaling with Multi-Sequence Verifiers

作者:
Yegon Kim↗Seungyoo Lee↗Chaeyun Jang↗Hyungi Lee↗Juho Lee↗

arXiv:2603.03417v2 Announce Type: replace-cross Abstract: Parallel test-time scaling, which generates multiple candidate solutions for a single problem, is a powerful technique for improving large language model performance. However, it is hindered by two key bottlenecks: accurately selecting the correct solution from the candidate pool, and the high inference latency from generating many full solutions. We argue that both challenges are fundamentally linked to verifier calibration, as a well-calibrated verifier improves answer selection and enables early-stopping strategies to reduce latency. However, existing non-generative verifiers are limited as they score each candidate in isolation, overlooking rich contextual information across the set of candidates. To address this, we introduce the Multi-Sequence Verifier (MSV), a lightweight verifier that predicts each candidate's correctness conditioned on the full sampled set. MSV achieves improved calibration, which directly enhances best-of-N selection performance and empowers a novel early-stopping framework. Across challenging mathematical reasoning benchmarks, MSV improves best-of-64 accuracy by up to 6\% relative to strong baselines, and in the early-stopping setting reaches the same accuracy as baselines with less than half the latency.

阅读与讨论 → 访问原文 →
04.
arXiv (CS.LG) 2026-06-16

In-Context Learning Is Provably Bayesian Inference: A Generalization Theory for Meta-Learning

作者:
Tomoya Wakayama↗Taiji Suzuki↗

arXiv:2510.10981v3 Announce Type: replace-cross Abstract: This paper develops a finite-sample statistical theory for in-context learning (ICL), analyzed within a meta-learning framework that accommodates mixtures of diverse task types. We introduce a principled risk decomposition that separates the total ICL risk into two orthogonal components: Bayes Gap and Posterior Variance. The Bayes Gap quantifies how well the trained model approximates the Bayes-optimal in-context predictor. For a uniform-attention Transformer, we derive a non-asymptotic upper bound on this gap, which explicitly clarifies the dependence on the number of pretraining prompts and their context length. The Posterior Variance is a model-independent risk representing the intrinsic task uncertainty. Our key finding is that this term is determined solely by the difficulty of the true underlying task, while the uncertainty arising from the task mixture vanishes exponentially fast with only a few in-context examples. Together, these results provide a unified view of ICL: the Transformer selects the optimal meta-algorithm during pretraining and rapidly converges to the optimal algorithm for the true task at test time.

阅读与讨论 → 访问原文 →
05.
arXiv (CS.CV) 2026-06-12

Emerging Flexible Designs for Geospatial Multimodal Foundation Models

作者:
Philipe Dias↗Waqwoya Abebe↗Abhishek Potnis↗Aristeidis Tsaris↗Dan Lu↗Xiao Wang↗Dalton Lunga↗

Foundation models are rapidly transforming Earth observation by enabling scalable pretraining across diverse unlabeled geospatial modalities. However, their architectural diversity ranging from encoder-only to encoder-decoder and masked autoencoding paradigms makes it challenging to assess performance trade offs in a consistent manner. In this work, we present an apples-to-apples comparison of leading FM architectures designed for geospatial multimodal reasoning, with a particular focus on flexibility across varied spectral band configurations. We standardize pretraining using identical self supervised learning objectives and training datasets, and evaluate all models under consistent parameterization on the GEOBench benchmark across classification and segmentation tasks. Our results offer new insights into the design trade-offs between model flexibility, modality alignment, and downstream task performance. By highlighting architectural strengths and limitations under controlled conditions, this study provides practical guidance for building next generation geospatial foundation models capable of robust multimodal reasoning.

阅读与讨论 → 访问原文 →
06.
arXiv (CS.CV) 2026-06-12

V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning

作者:
Lorenzo Mur-Labadia↗Matthew Muckley↗Amir Bar↗Mido Assran↗Koustuv Sinha↗Mike Rabbat↗Yann LeCun↗Nicolas Ballas↗Adrien Bardes↗

We present V-JEPA 2.1, a family of self-supervised models that learn dense, high-quality visual representations for both images and videos while retaining strong global scene understanding. The approach combines four key components. First, a dense predictive loss uses a masking-based objective in which both visible and masked tokens contribute to the training signal, encouraging explicit spatial and temporal grounding. Second, deep self-supervision applies the self-supervised objective hierarchically across multiple intermediate encoder layers to improve representation quality. Third, multi-modal tokenizers enable unified training across images and videos. Finally, the model benefits from effective scaling in both model capacity and training data. Together, these design choices produce representations that are spatially structured, semantically coherent, and temporally consistent. Empirically, V-JEPA 2.1 achieves state-of-the-art performance on several challenging benchmarks, including 7.71 mAP on Ego4D for short-term object-interaction anticipation and 40.8 Recall@5 on EPIC-KITCHENS for high-level action anticipation, as well as a 20-point improvement in real-robot grasping success rate over V-JEPA-2 AC. The model also demonstrates strong performance in robotic navigation (5.687 ATE on TartanDrive), depth estimation (0.307 RMSE on NYUv2 with a linear probe), and global recognition (77.7 on Something-Something-V2). These results show that V-JEPA 2.1 significantly advances the state of the art in dense visual understanding and world modeling.

阅读与讨论 → 访问原文 →
07.
arXiv (CS.CV) 2026-06-19

FrequencyFormer: A Co-Designed Sensor-to-Processor Pipeline for Frequency-Domain Vision Transformer Inference

作者:
Chengwei Zhou↗Ovishake Sen↗Xuming Chen↗Rishith Paramasivam↗Shaahin Angizi↗Swarup Bhunia↗Baibhab Chatterjee↗Gourav Datta↗

Deploying vision transformers (ViTs) on sensor-edge systems is limited not only by on-device compute, but also by the energy and bandwidth required to transmit high-dimensional image data from the sensor to the processor. While in-sensor and near-sensor computing reduce this cost through early feature extraction, existing methods often provide only modest compression. We observe that the frequency domain provides a naturally compact representation of visual information and can be exploited at the sensor level to reduce sensor-to-processor data movement. Building on this insight, we present FrequencyFormer, a co-designed sensor-to-processor pipeline for efficient ViT inference. FrequencyFormer includes: (1) a multi-scale DCT tokenizer that compresses a 224x224 image into compact frequency-domain tokens, achieving up to 128x reduction in off-chip data volume with modest accuracy loss; (2) a LUT-based near-sensor hardware implementation that leverages fixed DCT coefficients for multiplier-free, energy- and area-efficient tokenization; and (3) a modified MIPI-based low-power communication architecture that further reduces transfer energy. FrequencyFormer serves as a drop-in replacement for standard ViT patch embedding and remains compatible with pretrained backbones across classification, detection, and segmentation tasks. The pipeline achieves 28.8 TOPS/W, reduces communication energy by 230x, and lowers total sensor-side energy by 2.22x, demonstrating frequency-domain tokenization as a scalable foundation for in-sensor ViT deployment.

阅读与讨论 → 访问原文 →
08.
bioRxiv (Bioinfo) 2026-06-15

Biological meaning in protein embedding space is resolution-dependent

作者:
Zong↗Ren↗Li↗Finn↗R. D↗Wang↗

Protein language model embeddings are increasingly used to organise biological sequences, yet how biological meaning is encoded within embedding neighbourhoods remains poorly understood. Using two independent hierarchical enzyme systems, carbohydrate-active enzymes and peptidases, we investigated how biological interpretation changes across embedding organisations aligned to different levels of biological hierarchy. Different embedding organisations give rise to distinct neighbourhood semantics. When aligned to membership-boundary resolution, embeddings robustly separated artefacts and unrelated proteins from members of the target category. However, embeddings aligned to functional-grouping resolution maintained compositional neighbourhood structure for multi-domain proteins spanning more than one functional or catalytic group. Finally, embeddings aligned to local-family resolution recovered compact family-like neighbourhoods, including families withheld from training, while weakening broader membership-boundary and functional-grouping relationships. Moreover, embeddings optimised toward the same level of biological organisation retain different biological relationships depending on optimisation trajectory employed. Together, our results show that proximity in protein embedding space has no fixed biological interpretation. Instead, biological meaning emerges across embedding resolutions through selective preservation of different forms of biological organisation.

阅读与讨论 → 访问原文 →
09.
arXiv (CS.CL) 2026-06-19

Ensembles of Large Language Models for Identifying EQ-5D Studies in PubMed Based on Their Abstracts

作者:
Zhyar Rzgar K. Rostam↗M\'arta P\'entek↗J\'anos Tibor Czere↗Zsombor Zrubka↗L\'aszl\'o Gul\'acsi↗G\'abor Kert\'esz↗

The rapid increase in scientific publications leads to the fact that manual study screening in systematic literature reviews (SLRs) is increasingly resource consuming, inefficient, and inconsistent. Classifying studies that clearly report health-related quality-of-life results, such as EQ-5D data, requires a high level of clinical interpretation and poses challenges for human reviewers. This study investigates the use of Google's Gemini and Gemma large language models (LLMs) in automating EQ-5D detection in the PubMed biomedical database based only on published abstracts. A multi-phase framework is proposed that integrates few-shot prompting, weight ensembling aggregation, and a soft stacking meta-classifier. Nine LLMs are evaluated on a dataset of PubMed studies manually labeled by two experts regarding EQ-5D reporting. The weighted ensemble of gemini-2.5-pro, gemma-3-12b, and gemma-3-27b obtained a 0.74 weighted F1-score and 0.74 accuracy, exceeding individually attained results. The ensembling of top-performing models improved the balance between precision and recall compared to individual models, while the soft stacking approach provided greater reliability and interpretability. Feature analysis shows that the probability results from the models are important in guiding the final predictions. The findings suggest that an ensemble-based LLM setup is a reliable and scalable approach for automating screening in biomedical research.

阅读与讨论 → 访问原文 →
10.
arXiv (CS.AI) 2026-06-11

MetaPlate: Counterfactual-Guided RAG-LLM Tool for Personalized Food Recommendation and Hyperglycemia Prevention

作者:
Asiful Arefeen↗Carol Johnston↗Hassan Ghasemzadeh↗

arXiv:2606.10120v2 Announce Type: replace-cross Abstract: Postprandial hyperglycemia is a key risk factor for metabolic disorders; however, existing dietary guidance is often static, impractical, and insufficiently personalized, providing recommendations that are difficult to follow or not impactful. While recent advances leverage continuous glucose monitoring (CGM) and machine learning to predict glycemic responses, these approaches are largely predictive and lack actionable guidance. Moreover, recommendation systems are often misaligned with user goals and require extensive input. We present MetaPlate, a counterfactual explanation (CF) guided, context-aware decision-support framework that generates personalized meal recommendations to mitigate postprandial glucose excursions in healthy adults. MetaPlate integrates multimodal data, including CGM readings, wearable-derived physiological signals, and user-provided meal inputs from $25$ individuals to model pre-meal context. A machine learning model predicts glucose response, while a CF optimization module adjusts meal composition modifying macronutrient amounts to maintain glucose levels within a target range ($\leq 140$ mg/dL). An LLM-based retrieval-augmented generation (RAG) layer enhances interpretability by producing human-readable recommendations using constrained search of the USDA food database. We evaluate MetaPlate via a structured expert-in-the-loop assessment with registered dietitians (RDs), comparing performance before and after prompt refinement. Results show improvements in meal realism, portion suitability, and recommendation likelihood, with expert feedback indicating a shift from clinically implausible outputs to actionable, contextually appropriate recommendations. Our findings emphasize the importance of domain knowledge and structured constraints in LLM-driven systems and highlight the potential of MetaPlate as a real-time personalized dietary decision-support tool.

阅读与讨论 → 访问原文 →
11.
PLOS Medicine 2026-05-08

Climate change and non-communicable diseases: An invisible syndemic

作者:
Gokul Parameswaran↗

by Gokul Parameswaran, Sadeer Al-Kindi, Sanjay Rajagopalan Climate change accelerates non-communicable diseases (NCDs) through cascading environmental disruptions and is attributed to driving increased NCD-related mortality. Yet this syndemic remains invisible and underfunded. We detail why addressing the climate-NCD intersection is critical for improving health. In this Perspective, Sanjay Rajagopalan and colleagues discusses how climate change accelerates non-communicable diseases (NCDs) and exacerbates NCD-related mortality, and calls for greater visibility and funding to address this syndemic and improve human health.

阅读与讨论 → 访问原文 →
12.
arXiv (CS.AI) 2026-06-15

CisTransCell: Single-Cell Perturbation Prediction via Gene Function, Regulatory Control, and Cellular Context

作者:
Wei Zhang↗Xun Jiang↗Yuesi Xi↗Ming Tang↗

arXiv:2606.13713v1 Announce Type: cross Abstract: Predicting cellular transcriptional responses to genetic perturbations is a central problem in single-cell biology, especially in the zero-shot setting where the perturbed gene or gene combination is unseen during training. A major difficulty is that perturbation effects are not determined by expression state alone: they depend on how the perturbed gene product influences other genes and proteins, how those downstream factors act on cis-regulatory elements, and which regulatory programs are active in the current cell state. To better capture this biological complexity, we propose CisTransCell, a cell-conditioned multi-modal framework for single-cell perturbation prediction that augments each gene with two complementary priors: a regulatory-sequence prior that captures how the gene is controlled, and a coding-sequence prior that captures what the gene product does. By integrating these priors with cellular expression state, CisTransCell models perturbation response as a cascade from gene function to regulatory control to downstream transcriptional change. Experiments on benchmark single-cell perturbation datasets show that CisTransCell achieves strong performance in zero-shot perturbation prediction.

阅读与讨论 → 访问原文 →
14.
arXiv (CS.AI) 2026-06-15

Hidden in Plain Sight: Benchmarking Agent Safety Against Decomposition Attacks with DECOMPBENCH

作者:
Vikhyath Kothamasu↗Virginia Smith↗Chhavi Yadav↗

arXiv:2606.13994v1 Announce Type: cross Abstract: LLM-based Agents are becoming increasingly capable and widely deployed, creating growing incentives for adversarial misuse in the real-world. A key emerging threat is Decomposition Attacks [glukhov2024breach, jones2024adversaries] in which a harmful task is broken into simpler, benign subtasks that evade safety mechanisms when executed separately but cumulatively fulfill the malicious intent. Although recent benchmarks assess agent safety in multi-turn and multi-tool-use settings, they do not explicitly capture this form of decompositional misuse and may not represent realistic adversarial execution flows. To this end, we introduce DeCompBench, a benchmark designed specifically to evaluate agentic safety under decomposition attacks. DeCompBench is created with a decomposition-by-design principle using a graphical framework and enables harmful task decomposition into individually benign and executable subtasks with realistic workflows. Our experiments using a custom decomposer show that state-of-the-art agents exhibit high refusal rates on monolithic harmful tasks, but significantly lower refusal rates on their decomposed variants, while often inadvertently fulfilling the adversarial objectives. These findings underscore the need for safety evaluations against decomposition attacks and corresponding defenses. Our dataset is publicly available and can be found at https://huggingface.co/datasets/decompositionbench/DeCompBench.

阅读与讨论 → 访问原文 →
15.
arXiv (CS.CL) 2026-06-16

Evaluating the Robustness of Proof Autoformalization in Lean 4

作者:
Zhengtao Gui↗Sheng Yang↗Zhouxing Shi↗

Proof autoformalization aims to translate a mathematical informal proof written in natural language into a formal proof in a formal language such as Lean~4. Several works have developed LLM-based models for proof autoformalization. However, existing evaluations have typically focused on translating well-formed informal proofs from curated datasets. We argue that a robust proof autoformalizer must remain faithful even for informal proofs that diverge from these idealized ones, and we present the first study on the robustness of proof autoformalization models. We formulate two categories of perturbations and evaluate robustness under each: a global perturbation paraphrases the informal proof in a different style, under which the formalization should remain consistent; a local perturbation alters a value, symbol, or proof step, possibly in a counterfactual way, and a robust formalization should faithfully reflect the perturbation rather than reverting to the original one or inferring a different one on its own. We build a benchmark with both perturbations on miniF2F and MATH-500, and automatically measure how stable a proof autoformalization's correctness is under global perturbations and how faithfully its output reflects local perturbations. We evaluate seven recent models, all of which are sensitive to global perturbations and mostly fail to remain faithful under local perturbations. Code and data are available via https://github.com/ucr-rai/robust-proof-autoformalization.

阅读与讨论 → 访问原文 →
16.
arXiv (quant-ph) 2026-06-12

Exploring Exotic Spin-Dependent Interactions Beyond the Standard Model: Theoretical Foundations and Experimental Investigations

作者:
L. Y. Wu↗H. Yan↗

arXiv:2606.13318v1 Announce Type: cross Abstract: New interactions mediated by novel particles propose solutions to several important questions in modern physics. Axions serve as examples of such particles; they are lightweight and interact weakly with ordinary matter. This category of particles, including those similar to axions-termed Axion-Like Particles (ALPs)-arises from diverse theoretical frameworks, such as the Peccei-Quinn mechanism addressing the strong CP problem, string theory, and spontaneous supersymmetry breaking. Given their light mass and weak coupling, ALPs are also possible candidates for cold dark matter. Introducing these new interactions mediated by novel particles not only tackles several challenges in modern physics but also raises a crucial question: Are there undiscovered interactions beyond the Standard Model? Many of the interactions predicted by these theories are spin-dependent, which is the primary focus of this review. In this review, we first outline the theoretical foundations for investigating exotic spin-dependent interactions, highlighting their importance in various models beyond the Standard Model. We examine the potential roles of new lightweight particles in mediating these interactions, which may enhance our understanding of dark matter. Relevant formulas derived from theoretical models are included to support experimental investigations. Following this theoretical framework, we conduct a detailed review of recent experimental efforts to detect these exotic interactions. A systematic review of current constraints on these interactions is presented, along with an assessment of various detection approaches.

阅读与讨论 → 访问原文 →
17.
arXiv (CS.CV) 2026-06-17

A Benchmark for Omni-Modal Reasoning in Long Videos

作者:
Mohammed Irfan Kurpath↗Jaseel Muhammad Kaithakkodan↗Jinxing Zhou↗Sahal Shaji Mullappilly↗Mohammad Almansoori↗Noor Ahsan↗Beknur Kalmakhanbet↗Sambal Shikhar↗Rishabh Lalla↗Jean Lahoud↗Mariette Awad↗Fahad Shahbaz Khan↗…

Long-form omni-modal video understanding requires integrating vision, speech, and ambient audio with coherent long-context reasoning. Existing video benchmarks often trade off temporal scale, modality coverage, open-ended interaction, and interpretable scoring. To address this gap, we introduce LongShOTBench, a long video understanding benchmark designed around three coupled goals: holistic omni-modal integration, intent-driven open-ended interaction, and rubric-level diagnosis. It builds single- and multi-turn questions from real viewing scenarios, with systematic tasks probing visual, speech, ambient-audio, temporal, and cross-modal reasoning. Each item includes a reference answer and a weighted criterion-level rubric, letting evaluation identify which perceptual facts, temporal links, modality-grounding requirements, and reasoning steps are satisfied or missed. All samples are manually verified to improve grounding, clarity, and rubric reliability. We also introduce LongShOTAgent, a training-free omni-modal evidence-seeking agent coupling full-video preprocessing with targeted retrieval, query-adaptive segment refinement, and explicit claim verification over visual, speech, and non-speech audio evidence. Its iterative search-refine-verify loop exposes intermediate evidence and lets modality-specific specialists re-analyze relevant moments before answering. We evaluate 105 video-capable models spanning open-source omni-modal models, vision-language systems, audio LLMs, agentic pipelines and closed-source APIs. Current MLLMs remain far from saturating LongShOTBench, while our LongShOTAgent is the strongest training-free system, reaching 66.64% overall. By releasing the benchmark, leaderboard, and method, we provide a shared, interpretable testbed for advancing long-form omni-modal video reasoning. Code, data, and the leaderboard are available at https://longshot.cvmbzuai.com/.

阅读与讨论 → 访问原文 →
18.
arXiv (CS.AI) 2026-06-17

Dimensionality Controls When Modularity Helps in Continual Learning

作者:
Kathrin Korte↗Christian Medeiros Adriano↗Joachim Winther Pedersen↗Eleni Nisioti↗Sebastian Risi↗

arXiv:2606.17889v1 Announce Type: cross Abstract: Compositional learning systems must balance plasticity, the ability to acquire new knowledge, with stability, the preservation of previously learned components, especially when tasks share structure and risk interference. We study how modular architecture, task similarity, and representational dimensionality jointly shape compositional continual learning in a sequential A-B-A paradigm, comparing a task-partitioned recurrent network to a single-network baseline while inducing high- and low-dimensional regimes via weight-scale manipulations. In a high-dimensional "lazy" regime, both architectures achieve similar performance and internal geometry, suggesting that explicit modular structure has little impact when representations are weakly constrained. In a lower-dimensional "rich" regime, modularity becomes decisive: the modular network develops graded task-specific subspaces that overlap for similar tasks, partially align for moderately dissimilar tasks, and separate for dissimilar tasks, yielding a more compositional and interpretable organization than the single network. These findings identify the representational regime induced by initialization scale, which co-varies with representational dimensionality, as a key factor governing when compositional, modular structure is functionally beneficial in continual learning, and support viewing safety and robustness as problems of adaptive allocation of representational subspaces rather than fixed separation versus sharing.

阅读与讨论 → 访问原文 →
19.
arXiv (CS.CL) 2026-06-16

Rapid Poison: Practical Poisoning Attacks Against the Rapid Response Framework

作者:
David Huang↗Jaewon Chang↗Avidan Shah↗Prateek Mittal↗Chawin Sitawarin↗

The Rapid Response (RR) framework, deployed in production systems, including Anthropic's ASL-3 safeguards, continuously improves jailbreak-detection classifiers. When new jailbreaks emerge that bypass these classifiers, Rapid Response generates synthetic variants for training, helping the model generalize from the new attacks and quickly adapt. We reveal that prompt injection can infiltrate this pipeline to deliver poisoned samples into the classifier's training set, enabling two attack objectives: (I) targeted poisoning attacks that create false positives on harmless samples by categorizing them as a jailbreak, with a specific desired feature (e.g., certain formatting, subject, or keyword), (II) concept-based backdoor attacks that induce false negatives on jailbreak inputs, generalizing even to jailbreaks from attack strategies the defender explicitly trained against, when the backdoor trigger is present. Importantly, our threat model restricts adversaries to modifying only jailbreak samples (not benign data or labels), a constraint unexplored by prior work that makes the second objective particularly challenging. We address this with Omission Attack, which exploits a new phenomenon: when training on concept-absent unsafe samples, the classifier misassociates that concept's presence with the safe label. Both attacks cause substantial and in some cases near-complete label flipping at only a 1% poisoning rate, achieving up to 100% false positive rates and up to 96% false negative rates.

阅读与讨论 → 访问原文 →
20.
arXiv (CS.CL) 2026-06-11

Reassessing High-Performing LLMs on Polish Medical Exams: True Competence or Bias-Driven Performance?

作者:
Antoni Lasik↗Jakub Pokrywka↗{\L}ukasz Grzybowski↗Jeremi Ignacy Kaczmarek↗Gabriela Korza\'nska↗Janusz \'Swieczkowski-Feiz↗Oskar Pastuszek↗Paulina Hoffman↗Jakub Tomasz D\k{a}browski↗Wojciech Kusa↗

Large language models (LLMs) in medicine are mainly evaluated using multiple-choice question answering (MCQA), which can overestimate real clinical ability due to guessing strategies and answer biases. To address these limitations, we introduce an expanded and more challenging benchmark based on Polish medical exams, adding over 15,000 questions, two new domains, and four structural modifications that reduce MCQA-specific artifacts and better test reasoning. We evaluate 21 LLMs and show that evaluation design strongly affects results. Under our harder setup, the best model (Qwen3.5-122B) drops by 28.4 and 31 pp on English and Polish exams, respectively. Despite low evidence of data contamination, standard MCQA scores do not reliably reflect true medical competence. To facilitate further research, we make our benchmark publicly available.

阅读与讨论 → 访问原文 →
21.
arXiv (CS.AI) 2026-06-19

Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models

作者:
Hoang Phan↗Xianjun Yang↗Yuanshun Yao↗Jingyu Zhang↗Shengjie Bi↗Xiaocheng Tang↗Madian Khabsa↗Lijuan Liu↗Deren Lei↗

arXiv:2510.21978v2 Announce Type: replace-cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has delivered impressive gains in mathematical and multimodal reasoning and has become a standard post-training paradigm for contemporary language and vision-language models. However, the RLVR recipe introduces a significant risk of capability regression, in which models forget foundational skills after prolonged training without employing regularization strategies. We empirically confirm this concern, observing that open-source reasoning models suffer performance degradation on core capabilities such as perception and faithfulness. While imposing regularization terms like KL divergence can help prevent deviation from the base model, these terms are computed on the current task and therefore do not guarantee preservation of broader knowledge. Meanwhile, commonly used experience replay across heterogeneous domains makes it nontrivial to decide how much training emphasis each objective should receive. To address this, we propose RECAP-a replay strategy with dynamic objective reweighting for general knowledge preservation. Our reweighting mechanism adapts online using short-horizon signals of convergence and instability, shifting the post-training focus away from saturated objectives and toward underperforming or volatile ones. Our method is end-to-end and readily applicable to existing RLVR pipelines without training additional models or heavy tuning. Extensive experiments on benchmarks using Qwen2.5-VL-3B and Qwen2.5-VL-7B demonstrate the effectiveness of our method, which not only preserves general capabilities but also improves reasoning by enabling more flexible trade-offs among in-task rewards.

阅读与讨论 → 访问原文 →
22.
arXiv (CS.AI) 2026-06-11

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.AI) 2026-06-11

TreeSeeker: Tree-Structured Trial, Error, and Return in Deep Search

作者:
Zhuofan Shi↗Mingzhe Ma↗Lu Wang↗Fangkai Yang↗Pu Zhao↗Yiming Guan↗Youling Huang↗Wei Zhang↗Qingwei Lin↗Dongmei Zhang↗Saravan Rajmohan↗

arXiv:2606.11662v1 Announce Type: new Abstract: Deep search requires agents to answer complex questions through multi-step web search, browsing, evidence comparison, and synthesis. A central challenge is deciding how to search when several directions look plausible but only some will later lead to reliable evidence. If an agent greedily follows the current best-looking direction, it may keep extending a weak continuation. If it explores without discipline, it may waste budget on disconnected trials. We propose TreeSeeker, an inference-time framework for controlled trial-and-error in deep search. TreeSeeker organizes search as branch-and-return search over tree-structured states, where each branch is a tentative direction for a sub-goal. At each round, TreeSearch reads all sub-goal trees, identifies active goals, and uses textual UCB signals of value, uncertainty, and risk to select among exploiting a promising branch, exploring an uncertain alternative, or pruning an unproductive continuation and returning to an earlier branch point. TreeMem supports this control loop by keeping evidence, uncertainty, conflicts, progress, and failure cues attached to the branches that produced them, so trial outcomes can guide later decisions. Experiments on XBench-DeepSearch, BrowseComp, and BrowseComp-ZH show that TreeSeeker consistently outperforms strong open-source baselines, suggesting that explicit branch-and-return control complements stronger reasoning and tool execution.

阅读与讨论 → 访问原文 →
24.
arXiv (quant-ph) 2026-06-17

Unclonable Encryption in the Haar Random Oracle Model

作者:
James Bartusek↗Eli Goldin↗

arXiv:2603.11437v2 Announce Type: replace-cross Abstract: We construct unclonable encryption (UE) in the Haar random oracle model, where all parties have query access to $U,U^\dagger,U^*,U^T$ for a Haar random unitary $U$. Our scheme satisfies the standard notion of unclonable indistinguishability security, supports reuse of the secret key, and can encrypt arbitrary-length messages. That is, we give the first evidence that (reusable) UE, which requires computational assumptions, exists in "microcrypt", a world where one-way functions may not exist. As one of our central technical contributions, we build on the recently introduced path recording framework to prove a natural ``unitary reprogramming lemma'', which may be of independent interest.

阅读与讨论 → 访问原文 →
25.
arXiv (CS.CL) 2026-06-19

Quality Over Clicks: Iterative Reinforcement Learning for Early-Stage E-Commerce Query Suggestion

作者:
Qi Sun↗Kejun Xiao↗Huaipeng Zhao↗Tao Luo↗Xiaoyi Zeng↗

Existing dialogue systems rely on query suggestion to enhance user engagement. Recent approaches mainly optimize generative models using click-through rate (CTR) models to align with user preferences. However, these methods are less effective in early-stage deployment scenarios, where click feedback is sparse and insufficient for training a reliable CTR model. To bridge this gap, we propose QualEQS, a quality-first iterative reinforcement learning framework for e-commerce query suggestion. We formalize actionable suggestion quality along three dimensions that directly affect downstream usability: answerability, factuality, and information gain. To continuously improve from online traffic without click supervision, we further propose group-level disagreement among candidate suggestions to identify ambiguous query contexts and mine hard training cases for iterative refinement. We also introduce EQS-Benchmark, a dataset of 16,949 real-world e-commerce queries for offline training and evaluation. Experiments show that our quality-based offline metrics correlate strongly with online performance, providing a practical evaluation recipe for sparse-feedback deployment. In both offline and online settings, QualEQS consistently outperforms strong baselines, yielding a 6.81% improvement in online ChatPV in a real-world enterprise-level conversational shopping assistant system.

阅读与讨论 → 访问原文 →