论文广场 - AcademicHub

01.

arXiv (CS.CL) 2026-06-12 DOI: arXiv:2606.12413

AI SciBrief as a Gateway to Research: A Framework for Onboarding Students into New Research Areas

作者:

Students at all levels of higher education face a significant barrier in the form of information overload, which often paralyzes the initial stages of the research process and suppresses motivation. In response, this article introduces a pedagogical framework that leverages AI SciBrief, a platform powered by a Large Language Model (LLM) designed to automatically generate digests of scientific trends. We describe how this multidisciplinary tool - with initial coverage in finance, medicine, and education - can be integrated into the curriculum to overcome this "entry barrier." The framework provides concrete methodologies for utilizing these digests to facilitate topic selection for term papers, accelerate literature reviews for dissertations, and enable postgraduate students to continuously monitor emerging trends. We conclude that AI SciBrief functions as a "gateway to research" effectively reducing students' cognitive load and empowering them to transition more rapidly from information searching to knowledge creation.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2606.13995

Dialogue SWE-Bench: A Benchmark for Dialogue-Driven Coding Agents

作者:

Brendan King ↗Jeffrey Flanigan ↗

AI coding agents have rapidly transformed software engineering, powering widely used interactive coding assistants. Despite their interactive real-world use, existing benchmarks evaluate them as fully-autonomous systems. In this work, we introduce Dialogue SWE-Bench, an automatic benchmark dataset for evaluating the ability of coding agents to resolve real-world software engineering problems through dialogue with a user. We design a novel, persona-grounded user simulator to support our task evaluation, and augment our task evaluation with automatic evaluations of dialogue quality. We also propose a new schema-guided agent, aimed at improving the dialogue capabilities of off-the-shelf coding agents, which improves over strong baselines by 3-14%. Our results indicate that better coding models do not always correspond to better dialogue models, suggesting that dialogue capability is a distinct and currently understudied dimension of coding agent performance.

阅读与讨论 → 访问原文 →

03.

arXiv (quant-ph) 2026-06-17 DOI: arXiv:2606.18188

Learning Arbitrary Lindbladians with Quantum Error Correction

作者:

Nikita Romanov ↗Petr Ivashkov ↗Weiyuan Gong ↗Ishaan Kannan ↗Andi Gu ↗Hong-Ye Hu ↗Susanne F. Yelin ↗

arXiv:2606.18188v1 Announce Type: new Abstract: We study ansatz-free Lindbladian learning, the problem of reconstructing the generator of an open quantum system without prior knowledge of its Hamiltonian or dissipator structures. This problem exhibits two distinct information-theoretic precision limits: Hamiltonian components unmasked by dissipation are Heisenberg-limited, while the remaining Lindbladian components are subject to the quadratically worse standard quantum limit. Existing approaches that attain these optimal scalings strongly rely on pre-specified structure of interaction and noise, leaving the ansatz-free setting an open problem. In this work, we present the first standard-quantum-limited algorithm for learning arbitrary sparse Lindbladians. Under an additional physically motivated regularity condition, our framework also learns the Hamiltonian component disjoint from the dissipator at the Heisenberg limit, without prior knowledge of either the Hamiltonian or dissipator supports. Our main technical ingredient is a recursive random stabilizer-code construction that suppresses the strongest Lindbladian terms while preserving sensitivity to weaker unknown ones. These results establish a scalable framework for characterizing unknown open quantum systems, with quantum error correction serving as a key learning primitive.

阅读与讨论 → 访问原文 →

04.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.15019

Towards Global AI-Driven Cervical Cancer Screening

作者:

The global elimination of cervical cancer is a key public health goal set by the World Health Organization (WHO), with screening programs reducing mortality by up to 80%. However, access to experts and biopsy services is limited in low- to middle-income countries (LMICs). Deep learning (DL)-based algorithms offer promising support for screening, but most existing approaches have been developed and validated on private datasets from single countries. We present the first DL-based approach to cervical cancer screening validated on data from multiple countries. Technically, we phrase the problem of detecting and classifying lesions in colposcopy images as a multi-task learning problem, in which we simultaneously perform image-level classification and lesion segmentation. Our model was trained on a private data set of acid stain colposcopy images with manually generated lesion segmentation masks and corresponding histopathological results, employing extensive data augmentation to address image variability. In an in-distribution validation with pathology results serving as ground truth, our algorithm outperformed medical experts (Balanced Accuracy: 0.68 vs 0.64) in CIN1- (Cervical intraepithelial neoplasia grade 1 or lower) versus CIN2+ (grade 2 or higher) classification. External validation on four colposcopy data sets from four countries featuring radical differences in prevalence and patient characteristics yielded superior performance of our method compared to baseline methods. Performance variability across countries was high with AUC values ranging from 0.54 - 0.80. Overall, algorithm performance varied with age, transformation zone (cervical area most prone to lesion development), presence of comorbidities and pathognomonic signs, with comorbidities having by far the largest negative effect. Future work should focus on improving model robustness and generalizability.

阅读与讨论 → 访问原文 →

05.

arXiv (CS.AI) 2026-06-12 DOI: arXiv:2602.07294

Fin-RATE: A Real-world Financial Analytics and Tracking Evaluation Benchmark for LLMs on SEC Filings

作者:

Yidong Jiang ↗Junrong Chen ↗Eftychia Makri ↗Jialin Chen ↗Peiwen Li ↗Ali Maatouk ↗Leandros Tassiulas ↗Eliot Brenner ↗Bing Xiang ↗Rex Ying ↗

arXiv:2602.07294v4 Announce Type: replace-cross Abstract: With the increasing deployment of Large Language Models (LLMs) in the finance domain, LLMs are increasingly expected to parse complex regulatory disclosures. However, existing benchmarks often focus on isolated details, failing to reflect the complexity of professional analysis that requires synthesizing information across multiple documents, reporting periods, and corporate entities. Furthermore, these benchmarks do not disentangle whether errors arise from retrieval failures, generation inaccuracies, domain-specific reasoning mistakes, or misinterpretation of the query or context, making it difficult to precisely diagnose performance bottlenecks. To bridge these gaps, we introduce Fin-RATE, a benchmark built on U.S. Securities and Exchange Commission (SEC) filings and mirroring financial analyst workflows through three pathways: detail-oriented reasoning within individual disclosures, cross-entity comparison under shared topics, and longitudinal tracking of the same firm across reporting periods. We benchmark 17 leading LLMs, spanning open-source, closed-source, and finance-specialized models, under both ground-truth context and retrieval-augmented settings. Results show substantial performance degradation, with accuracy dropping by 18.60% and 14.35% as tasks shift from single-document reasoning to longitudinal and cross-entity analysis. This degradation is associated with increased comparison hallucinations, temporal and entity mismatches, and is further reflected in declines in reasoning quality and factual consistency–limitations that existing benchmarks have yet to formally categorize or quantify.

阅读与讨论 → 访问原文 →

06.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2606.15427

Post-Launch Capability Expansion of Vision-Language Models via Prompting for On-Orbit Spacecraft Inspection

作者:

Nicholas A. Welsh ↗Lennon J. Shikhman ↗Monty Nehru Attazs ↗Seemanthini K. Putane ↗Van Minh Nguyen ↗Ryan T. White ↗

Spaceborne inspection systems often deploy perception models prior to launch, after which updating model weights or expanding fixed label sets becomes operationally impractical. While supervised models can be integrated pre-flight, adding new semantic capabilities in orbit requires retraining and re-uploading parameters. We investigate whether prompt-driven vision–language models can enable post-launch semantic expansion, allowing new spacecraft components to be specified via natural-language prompts without modifying onboard weights. We evaluate zero-shot instance segmentation of spacecraft components under a strictly frozen, single-pass inference protocol on a test set of $129$ images of previously unseen satellites. Under fixed global thresholds and no post-processing, SAM3 achieves $0.385$ mAP@$0.5$ and $0.267$ mAP@$0.5{:}0.95$. Performance is strongly scale-dependent: large structural elements like spacecraft bodies ($0.639$ AP@$0.50$) and solar arrays ($0.598$ AP@$0.5$) localize reliably, while relatively small appendages like antennas ($0.221$ AP@$0.5$) and thrusters ($0.081$ AP@$0.5$) remain difficult. Prompt formulation influences performance, with structured prompts incorporating spatial and geometric descriptors yielding up to $82%$ improvement over short category-name prompts. The model operates within the memory and compute envelope of contemporary embedded GPUs, suggesting prompt-driven grounding can provide a practical mechanism for post-launch semantic extension of dominant spacecraft structures while highlighting limitations of zero-shot localization for fine-scale components under orbital domain shift.

阅读与讨论 → 访问原文 →

07.

arXiv (CS.CV) 2026-06-15 DOI: arXiv:2606.14072

Diffusion-Refined Segmentation and Vision-Language Interpretation for Pediatric Brain Tumor MRI

作者:

Wentao Ke ↗Jianche Liu ↗

Accurate pediatric brain tumor segmentation remains challenging due to limited annotated data, heterogeneous imaging phenotypes, diffuse tumor boundaries, and class imbalance across tumor subregions. Here, we present a two-stage deep learning framework for improving multi-modal pediatric brain MRI segmentation and clinical interpretation. First, we evaluate 3D Res U-Net and Swin-UNETR baselines on BraTS-PEDs MRI scans, using four co-registered modalities to predict tumor core, whole tumor, and enhancing tumor regions. Second, we introduce diffusion-based refinement models conditioned on coarse Swin-UNETR predictions, including a 3D DDPM refiner and MedSegDiff. Conditioning substantially improves diffusion stability and performance, particularly for enhancing tumor boundary segmentation. Conditioned MedSegDiff achieves the strongest boundary agreement with the lowest HD95. Finally, predicted tumor volumes and representative segmentation overlays are integrated with a multimodal language model to generate structured radiology-style reports. Together, our results suggest that coarse-to-refined diffusion segmentation can improve pediatric tumor boundary delineation and support end-to-end interpretable AI-assisted neuro-oncology workflows.

阅读与讨论 → 访问原文 →

08.

arXiv (CS.CL) 2026-06-11 DOI: arXiv:2606.12397

Redesign Mixture-of-Experts Routers with Manifold Power Iteration

作者:

Songhao Wu ↗Ang Lv ↗Ruobing Xie ↗Yankai Lin ↗

Router is the cornerstone component to the Mixture-of-Experts models. Serving as expert proxies, the rows of the router matrix compute their similarity to the MoE inputs to determine which subset of experts is activated. Ideally, each router row is designed to encode the expert matrix into this representative vector, such that its dot-product with token can better reflect token-expert affinity. However, there exists no design principles to enforce this condensation. In this paper, we propose to align each router row with the principal singular direction of the associated expert, as this direction provides the most expressive mathematical description of a matrix. Based on this principle, we propose a router redesign with Manifold Power Iteration (MPI). Specifically, it introduces a "Power-then-Retract" paradigm, where a power iteration step is performed on the router weights, followed by a retraction to impose a norm constraint to ensure both efficiency and stability. Theoretically, we show that MPI drives router rows to converge toward the principal singular directions of associated experts. Empirically, we pretrain MoE model across scales from 1B to 11B parameters to confirm that this alignment facilitates more effective MoE models.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.CV) 2026-06-18 DOI: arXiv:2606.18721

Rethinking the Pointer Loss in Table Structure Recognition: Geometry-Aware Pointer Loss for Spatial Locality

作者:

Hong-Jun Choi ↗Jongho Lee ↗Jaeyoung Kim ↗

Table Structure Recognition (TSR) using a pointer network achieves impressive results by predicting HTML sequences while aligning tags to detected text (or cell) regions. However, our analysis reveals that when pointer networks fail, 79.6% of errors occur between spatially adjacent cells (Manhattan distance

阅读与讨论 → 访问原文 →

10.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2606.20502

Calibration Without Comprehension: Diagnosing the Limits of Fine-Tuning LLMs for Vulnerability Detection in Systems Software

作者:

Arastoo Zibaeirad ↗Marco Vieira ↗

arXiv:2606.20502v1 Announce Type: cross Abstract: Whether LLMs scoring well on vulnerability benchmarks genuinely reason about security or merely pattern-match on contaminated data remains unresolved. We present CWE-Trace, a framework for LLM vulnerability detection built from 834 manually curated Linux kernel samples spanning 74 CWEs. The framework enforces a strict temporal split (pre-2025 historical set / post-cutoff leakage-free set), preserves context-aware vulnerable–patched pairs, and introduces two diagnostic metrics: the Directional Failure Index (DFI) and Hierarchical Distance and Direction (HDD). We evaluate eight vanilla LLMs and 15 LoRA fine-tuned variants across non-targeted detection, targeted detection, and CWE classification. Our analysis yields two key results. First, data contamination provides no measurable advantage. Function-level analysis shows that 84% of nominally contaminated samples carry no usable memorization signal: vulnerable functions are absent or cross-mapped across datasets, and ~31% of contaminated samples carry CWE misclassification. Second, backbone directional priors dominate fine-tuning. Models exhibit stable, systematic failure modes (DFI ranging from -85.5 to +94.8 pp) that persist from historical to post-cutoff data and resist correction. Fine-tuning shifts the output threshold without changing the decision policy. This is calibration without comprehension: output distributions adapt to training data while the underlying security reasoning remains absent. The weakest backbone at binary detection (DeepSeek-R1) gains the most in coarse CWE classification, revealing that detection and understanding are decoupled capabilities. The best detection score reaches only 52.1% (+2.1 pp above chance); exact CWE ranking remains below 1.3% Top-1 accuracy, confirming that current LLMs lack reliable security reasoning for systems software, regardless of fine-tuning strategy.

阅读与讨论 → 访问原文 →

11.

arXiv (CS.LG) 2026-06-15 DOI: arXiv:2606.14398

A theoretical model for task routing in mixture-of-expert transformers

作者:

Yongli Xiang ↗Vinoth Nandakumar ↗Yunzhi Yao ↗Peike Li ↗Tongliang Liu ↗

arXiv:2606.14398v1 Announce Type: new Abstract: Mixture-of-experts (MoE) layers enable the scaling of transformer models while keeping the inference compute fixed. While task-expert specialization has been observed in empirical studies of frontier MoE transformer models, existing theoretical work analyzes this using continuous mixture models that cannot be used to model natural language effectively. An important open question is to theoretically explain task-expert specialization in transformer MoE models using discrete models of language. To address this, we represent structured knowledge via syntactic templates and finite key-value dictionaries, and prove formally that a single-layer MoE transformer can encode knowledge by using experts that specialize in the corresponding tasks. Our construction shows how queries are routed to unique, task-specific experts whose size depends solely on the intrinsic complexity of the given task (i.e. the combined size of its syntactic templates and factual dictionary). Our construction provides a theoretical support for empirical results on localized knowledge circuits in MoE models. We support our theoretical findings with experiments evaluating model performance under varying MoE loss functions.

阅读与讨论 → 访问原文 →

12.

medRxiv (Medicine) 2026-06-18 DOI: HASH:b89114aa8bc21eafdfb003ad7ec79312

Maternal and fetal HLA heterozygosity in preeclampsia: Insights from a large multi-ancestry pregnancy cohort

作者:

Cao ↗Maher ↗Hu ↗Keating ↗B. J ↗Burwick ↗R. M ↗Karumanchi ↗S. A ↗Maxwell ↗G. L ↗Powe ↗…

Preeclampsia (PE) is a leading cause of maternal and neonatal morbidity, with immune dysregulation at the maternal-fetal interface central to its pathogenesis. The highly polymorphic human leukocyte antigen (HLA) region mediates maternal immune tolerance of the semi-allogeneic fetus, yet the contribution of HLA diversity to PE risk remains poorly defined. Whether the HLA heterozygote advantage observed in other immune disorders is relevant to PE has not been systematically evaluated. Using data from the multi-ancestry TOPMed Boston-Colombia Collaborative for Adverse Pregnancy Outcomes (n = 12,790; 4,770 PE, 8,020 controls; 10,808 maternal, 1,982 fetal, including 1,848 pairs), we evaluated associations between heterozygosity across eight classical HLA loci and PE and four sub-phenotypes, adjusting for genetic ancestry. HLA heterozygosity was common across most loci (>80%). No individual maternal HLA locus was associated with overall PE; however, heterozygosity across class I loci showed a protective effect in preterm PE (OR=0.82, 95%CI:0.69-0.97), with a similar pattern for HLA-A heterozygosity (OR=0.78, 95%CI:0.64-0.96). In contrast, fetal heterozygosity at HLA-DQB1 was nominally associated with increased risk of PE (OR=1.36, 95%CI:1.03-1.79) and preterm PE (OR=1.73, 95%CI:1.13-2.73). No individual maternal or fetal HLA alleles were associated with PE. Maternal-fetal mismatch analysis demonstrated locus-specific associations with preterm PE, including increased risk with HLA-DQA1 mismatch and reduced risk with HLA-C mismatch. These findings highlight distinct maternal and fetal immunogenetic contributions to PE risk and underscore the importance of considering HLA diversity-rather than individual alleles alone-in studies of PE etiology.

阅读与讨论 → 访问原文 →

13.

PLOS Computational Biology 2026-06-02 DOI: HASH:6188f77dbed607e9c3003c1811b3e459

Data-driven model reveals increased stability of CAG-expanded <i>huntingtin</i> RNA due to MID1 binding

作者:

Yuhong Liu ↗

by Yuhong Liu, Annika Reisbitzer, Domagoj Dorešić, Jan Hasenauer, Sybille Krauß, Tatjana Tchumatchenko RNA-binding proteins (RBP) are important regulators of RNA metabolism. In neurodegenerative disorders such as Huntington’s Disease (HD), disrupted RBP-RNA interactions contribute to neuronal dysfunction. One such RBP, Midline 1 (MID1), has been shown to aberrantly associate with mutant huntingtin (Htt) RNA, enhancing its translation, yet the mechanism driving this effect remains unknown. Here, we develop a computational model to understand the role of MID1. Based on previously published data, our model predicts that MID1 increases the stability of the Htt RNA. We experimentally validate this prediction, showing that overexpression of MID1 significantly prolongs the half-life of mutant Htt RNA. Furthermore, we evaluate model refinements, including clustering of MID1-bound RNA, which allow capturing all key observations in the data. Together, we provide a data-driven framework that underlines the importance of RBP-RNA interaction in post-transcriptional regulation. This framework also shows how individual molecular reactions jointly determine RNA stability and protein levels in HD.

阅读与讨论 → 访问原文 →

14.

arXiv (CS.AI) 2026-06-12 DOI: arXiv:2606.12703

SMSR: Certified Defence Against Runtime Memory Poisoning in Persistent LLM Agent Systems

作者:

Tarun Sharma ↗

arXiv:2606.12703v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) agents increasingly run with persistent memory that accumulates across user sessions. This creates a new attack surface: an adversary interacting only through normal channels can inject crafted memories that, once retrieved, steer the agent's responses for future users, without touching model weights or code. We call this Multi-Session Memory Poisoning (MSMP) and show that no existing defence certifies against it; static-corpus defences (RobustRAG, ReliabilityRAG) assume a fixed knowledge base, and heuristic filters are bypassed by fluent enterprise-style text. We present Signed Memory with Smoothed Retrieval (SMSR), the first defence with a certified robustness bound for this setting. Component 1 adds HMAC-SHA256 provenance at write time, blocking unsigned injection. Component 2 applies randomised memory ablation with verdict-based majority voting at query time, bounding the influence of authenticated adversaries. We prove that no provenance-free retrieval-time filter can certify against adaptive injection, derive a hypergeometric certificate for Component 2, and formalise the Consistent Minority Effect, whereby a consistent adversarial answer wins string-based voting as a numerical minority while verdict-based voting removes it. Across 15 enterprise scenarios (3,150 repeated trials), Component 1 cuts attack success from 93-100% to 0% for all unsigned variants. For an authenticated adversary with a single injection, Component 2 holds success to 8.0% (95% CI [5.8, 10.9], n=450), below the certified worst case. In an end-to-end query-only attack where the agent itself writes the poison rather than it being pre-seeded, SMSR reduces success from 65.3% to 5.3% (n=150, non-overlapping CIs) on a live agent stack. Clean-query utility is 90% (Component 1) and 85% (combined).

阅读与讨论 → 访问原文 →

15.

arXiv (CS.CL) 2026-06-17 DOI: arXiv:2606.17113

The Critical Role of Model Selection in Causal Inference: A Comparative Analysis of Classification Models within the InferBERT Framework for Pharmacovigilance

作者:

Csaba Kiss ↗Roland Molontay ↗Gabriele Pergola ↗

Distinguishing causal adverse drug events (ADEs) from spurious correlations remains a central challenge in pharmacovigilance. The InferBERT framework integrates transformer models with Do-calculus, but its success hinges on the underlying classification model. This study evaluates the impact of model choice in InferBERT, assessing whether simpler models suffice, if domain-specific pre-training helps, whether scaling to LLMs improves causal detection, and the effect of post-hoc calibration. We performed a comparative study on two benchmarks: Analgesics-induced Acute Liver Failure (AILF) and Tramadol-related Mortalities (TRAM). Four models were evaluated-XGBoost (baseline), ALBERT (original InferBERT), BioBERT (biomedical transformer), and Med-LLaMA (medical LLM)-using 5-fold cross-validation repeated over 20 runs. We measured accuracy, Expected Calibration Error (ECE) pre- and post-isotonic regression, and Jaccard concordance of causal terms with PRR, ROR, and EBGM; significance was tested with paired t-tests. BioBERT achieved the highest accuracy on both datasets, while Med-LLaMA underperformed despite its size and parameter-efficient fine-tuning. Domain-specific pre-training was decisive. Calibration improved ECE but had mixed effects on accuracy and causal discovery. BioBERT's superiority also yielded the strongest concordance with traditional pharmacovigilance signals. These results show that domain-specific pre-training provides a clear advantage over simpler baselines and larger LLMs. Investing in manageable, domain-aware models is more effective for computational pharmacovigilance than simply scaling model size.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2606.20376

CRAX: Fast Safe Reinforcement Learning Benchmarking

作者:

Tristan Tomilin ↗Mourad Boustani ↗Mickey Beurskens ↗Thiago D. Sim\~ao ↗

arXiv:2606.20376v1 Announce Type: cross Abstract: Safety is a core concern for deploying reinforcement learning (RL) agents in real-world domains such as robotics and autonomous driving. While benchmarks have been central to progress in RL, existing safety benchmarks with high-fidelity 3D physics remain computationally slow, limiting large-scale experimentation and rapid prototyping. To address this gap, we propose CRAX (Constrained RL Accelerated with JAX). Built on top of the MuJoCo XLA (MJX) physics engine with realistic 3D dynamics, CRAX leverages vectorized operations and hardware acceleration, yielding up to ~100x speedups over comparable CPU-based safety benchmarks. The benchmark features six environment suites and three agent-specific tasks, each spanning three difficulty levels. Evaluating six popular safe RL methods shows that no single approach dominates across all tasks, and reveals the trade-offs between performance and safety. We find that curriculum learning across difficulty levels and safety transfer can improve performance over direct training in harder settings.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2605.22664

MBABench: Evaluating LLM Agents on End-to-End Spreadsheet Tasks in Finance

作者:

Thomson Yen ↗Julian Poeltl ↗Harshith Srinivas Gear ↗Yilin Meng ↗Joshua Fan ↗Adam Shen ↗Yili Liu ↗Ali Bauyrzhan ↗Siri Du ↗Haoyang Liu ↗Daniel Guetta ↗Hongseok Namkoong ↗…

arXiv:2605.22664v3 Announce Type: replace Abstract: LLM agents are increasingly expected to carry out end-to-end workflows, producing complete artifacts from high-level user instructions. To meet enterprise needs, frontier AI labs have developed agents that can construct entire spreadsheets from scratch. This is especially relevant in finance, where core workflows such as financial modeling, forecasting, and scenario analysis are commonly conducted through spreadsheets. Yet, existing spreadsheet benchmarks do not measure this advanced capability, focusing instead on question-answering or single-formula edits. To address this gap, we provide one of the first evaluations of agents on end-to-end spreadsheet tasks, focusing on economically critical financial workflows such as modeling and scenario analysis. Since deliverables therein are routinely reviewed and revised by multiple stakeholders, judging their quality necessarily involves high-level criteria such as readability or ease of modification. To reflect the multidimensional nature of solution quality, we develop an evaluation taxonomy comprising three dimensions: Accuracy, Formula, and Format, each comprising fine-grained criteria that reflect professional standards. The Claude family leads the benchmark and produces the most professional-looking outputs in our qualitative review, but even the strongest agents frequently fall short of professional finance standards and degrade sharply as the difficulty increases beyond a few chained calculations. This suggests that current agents are not yet able to reliably produce professional-quality spreadsheets at the level of complexity real-world workflows demand.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2505.23277

Sentinel: Decoding Context Utilization via Attention Probing for Efficient LLM Context Compression

作者:

Yong Zhang ↗Heng Li ↗Yanwen Huang ↗Ning Cheng ↗Yang Guo ↗Yun Zhu ↗Yanmeng Wang ↗Shaojun Wang ↗Jing Xiao ↗

Retrieval-augmented generation (RAG) often suffers from long and noisy retrieved contexts. Existing context compression methods typically rely on heuristic relevance estimation or supervised compression models rather than on how LLMs utilize retrieved context during inference. We propose Sentinel, a lightweight sentence-level compression framework that decodes inference-time contextual utilization behaviors from head-wise attention patterns of frozen LLMs. To ground supervision in retrieval-dependent answering behavior, Sentinel trains a lightweight probe using QA examples where the model succeeds only when retrieved context is available. Sentinel performs compression using only a single non-autoregressive forward pass without dedicated compression training or autoregressive scoring. Empirically, we find that effective contextual utilization signals remain accessible even in compact proxy models. On LongBench, Sentinel with a 0.5B proxy model achieves up to 5$\times$ compression while attaining question-answering performance competitive with compression methods built on 7B-scale models. Despite being trained only on English QA data, Sentinel also generalizes effectively to Chinese and out-of-domain settings.

阅读与讨论 → 访问原文 →

19.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2606.11889

Task-Aligned Stability Analysis of Vision-Language Models for Autonomous Driving Hazard Detection

作者:

Everett Richards ↗

Vision-language models (VLMs) are increasingly used for scene understanding in autonomous driving, but robustness analysis often relies on task-agnostic embedding stability alone. We study whether corruption-induced embedding drift predicts changes in a task-aligned hazard score derived from CLIP image-text similarities. Using controlled corruptions on BDD100K road scenes, we compare embedding drift against margin drift, defined as the change in hazard score under perturbation. The relationship is highly corruption-dependent: some families exhibit strong coupling between representation drift and decision drift, while others induce hazardous decision instability despite relatively modest embedding change. Furthermore, corruption families differ in failure direction: most suppress hazard detections via false negatives, while occlusion instead triggers false alarms, suggesting that benchmark design should account for asymmetric failure modes, not just overall instability rates. These results suggest that robustness benchmarks should include task-aligned stability measures in addition to embedding-level perturbation statistics.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.AI) 2026-06-19 DOI: arXiv:2605.22748

Superhuman Safe and Agile Racing through Multi-Agent Reinforcement Learning

作者:

Ismail Geles ↗Leonard Bauersfeld ↗Markus Wulfmeier ↗Davide Scaramuzza ↗

arXiv:2605.22748v2 Announce Type: replace-cross Abstract: Autonomous systems have achieved superhuman performance in isolation or simulation, yet they remain brittle in shared, dynamic real-world spaces. This failure stems from the dominant single-agent paradigm for physical applications, where other actors are ignored or treated as environmental noise, preventing effective coordination. Here we show that multi-agent reinforcement learning provides the essential safety scaffolding required for real-world interaction. Using high-speed quadrotor racing as a high-stakes testbed, we train agents to navigate complex aerodynamic interactions and strategic maneuvering with a variable number of racers. Through league-based self-play, agents evolve sophisticated anticipatory behaviors, including proactive collision avoidance, overtaking, and handling multi-agent physical interactions, including aerodynamic downwash. Our agents outperform a champion-level human pilot in multi-player races at speeds exceeding 22 m/s, while simultaneously reducing collision rates by 50 % compared to state-of-the-art single-agent baselines. Crucially, training with diverse artificial agents enables zero-shot generalization to safer human interaction. These results suggest that the path to robust robotic co-existence lies not in isolated safety constraints, but in the rigorous demands of multi-agent interaction. Multimedia materials are available at: https://rpg.ifi.uzh.ch/marl

阅读与讨论 → 访问原文 →

21.

PLOS Medicine 2026-05-29 DOI: HASH:ce33ca594d10abc18dde2823a4b8a3eb

Availability, appeal, and addictiveness by design: Tobacco and nicotine industry deliberate targeting of youth

作者:

Raglan Maddox ↗

by Raglan Maddox, Becky Freeman, Charlotta Pisinger, Emily Banks Contemporary tobacco and nicotine products, particularly e-cigarettes, are deliberately designed, marketed, and distributed to maximize youth appeal, uptake, dependence, and use. Youth uptake is a predictable outcome of systems designed to maximize product availability, appeal, and addictiveness. In recognition of the World No Tobacco Day 2026 theme, "unmasking the appeal", this Perspective by Raglan Maddox and colleagues discusses how tobacco and nicotine products, particularly e-cigarettes, are deliberately designed and marketed to maximize youth appeal, and highlight the need for policies to ensure greater industry accountability and to tackle concerning uptake trends.

阅读与讨论 → 访问原文 →

22.

arXiv (CS.CL) 2026-06-15 DOI: arXiv:2606.13931

DLawBench: Evaluating LLMs Through Multi-Turn Legal Consultation

作者:

Li Zhang ↗Yuzhen Shi ↗Yiran Hu ↗Jingwen Zhang ↗Wenbo Lv ↗Yubo Ma ↗Wei Wang ↗Rongyao Shi ↗Yuanyang Qiu ↗Xinran Xu ↗Yuemeng Qi ↗Linlin Miao ↗…

Lawyer-client consultation is a critical starting point for legal services. Effective legal assistance hinges on eliciting sufficient and truthful information from clients in order to devise strategies that best protect their interests. This task requires Large Language Models (LLMs) not only to perform robust legal reasoning, but also to strategically elicit material facts through multi-turn interactions and effectively guide clients with diverse personalities. Yet existing legal benchmarks overlook this interactive capability. To fill this gap, we introduce DLawBench, a diagnostic benchmark for real-world legal consultation. Drawing on realistic client behavior, we characterize lawyer-client interactions into four types: Cooperative, Dependent, Withdrawn, and Adversarial. Using dialogues grounded in real cases, DLawBench evaluates whether LLMs can effectively conduct legal consultation under realistic conditions. DLawBench comprises 461 cases from Chinese and U.S. law, 5,532 paired fact entries, 3,411 inquiry rubrics, and 3,348 issue-resolution rubrics, and evaluates 26 representative LLMs. Systematic experiments show substantial headroom: the best-performing model, GPT-5.5, achieves only 0.562 on consultation-grounded legal reasoning. More importantly, DLawBench exposes both sycophancy in legal consultation and a paradox: models perform worse when clients need guidance most.

阅读与讨论 → 访问原文 →

23.

arXiv (CS.LG) 2026-06-18 DOI: arXiv:2606.18464

Modeling Doppler Shifts in Radial-Velocity Data with Deep Learning toward Earth-mass Exoplanet Detection

作者:

Isidro G\'omez-Vargas ↗Xavier Dumusque ↗Yinan Zhao ↗Khaled Al Moulla ↗Michael Cretignier ↗

arXiv:2606.18464v1 Announce Type: cross Abstract: Detecting the tiny Doppler shifts induced by Earth-mass planets in stellar radial-velocity measurements remains extremely challenging due to stellar activity. Many deep-learning methods performing well on simulated data remain difficult to apply reliably on real stellar spectra. The aim of this work is to develop a deep-learning framework that generalizes to real, unseen spectra and improves the detectability of Earth-mass planets in radial-velocity data. We train artificial neural networks on HARPS-N solar spectra with injected planetary signals, using physics-motivated spectral representations based on flux and line-formation temperature, together with their velocity gradients. Two training strategies are explored: hold-out testing and cross-validation. Model robustness is enhanced through genetic-algorithm-based hyperparameter optimization, and predictive uncertainty is quantified using Monte Carlo dropout. Our most precise neural network model reliably retrieves, under the cross-validation strategy, the amplitudes, phases, and orbital periods of planetary signals with amplitudes greater than or equal to 25 cm/s and periods between 10 and 550 days. In addition, in all cases tested here, the successfully recovered signals correspond to the most significant peaks in the periodograms of the Doppler-shift predictions. Temperature-based spectral-shell representations consistently outperform flux-based shells. We also release doppleriann, a Python package implementing the proposed framework. Our results demonstrate that combining physically motivated spectral representations with deep learning provides a promising pathway toward the detection of Earth-mass planets in radial-velocity data from real observations, supported by a modeling framework that is both physically grounded and statistically rigorous, incorporating uncertainty quantification and optimized training strategies.

阅读与讨论 → 访问原文 →

24.

arXiv (quant-ph) 2026-06-16 DOI: arXiv:2606.14825

Experimental realization of the complete seven-phase Anderson-localization landscape

作者:

Yao Qin ↗Chao Yang ↗Yuzhe Zhang ↗Yucheng Wang ↗Jingyun Fan ↗

arXiv:2606.14825v1 Announce Type: cross Abstract: Anderson localization has evolved far beyond the conventional dichotomy between extended and localized states. Modern localization theory predicts a complete transport hierarchy comprising extended, critical, and localized phases together with all coexistence phases among them, forming a seven-phase Anderson-localization landscape. Despite its fundamental importance, this hierarchy has never been experimentally realized within a single system. Here we realize the complete seven-phase Anderson-localization landscape in a one-dimensional Floquet photonic lattice. By engineering quasiperiodic hopping profiles containing inhomogeneously distributed hopping zeros, we generate critical states and enable their coexistence with extended and localized sectors. The resulting transport regimes are directly resolved through their distinct spatiotemporal dynamics, including ballistic expansion, confined critical oscillations, and persistent localization. We observe all seven phases, including the elusive triply coexisting extended-critical-localized phase, and experimentally track the phase transitions connecting them. Our results establish the first complete experimental map of the Anderson-localization landscape and provide a unified platform for investigating mobility edges, multifractality, and programmable coherent transport.

阅读与讨论 → 访问原文 →

25.

PLOS Computational Biology 2026-06-12 DOI: HASH:5c8c918a8a7100b969116a452c936bdc

Ten simple rules for executing an inherited research plan in computational biology

作者:

Sahar Javaheri Tehrani ↗

by Sahar Javaheri Tehrani, Toni Ingolf Gossmann Trainees in computational biology frequently inherit research plans whose aims, datasets, analytical strategies, and technical constraints were defined before their arrival. These plans often emerge from grants, collaborations, legacy codebases, shared high-performance computing environments, or partially completed analyses. While such plans provide a useful scaffold, they rarely specify all implementation details, prior assumptions, evaluation criteria, or dependencies needed for reliable execution. The transition from inheriting a partially articulated plan to producing reproducible results therefore creates an execution gap: a phase in which trainees must reconstruct what the project is, which elements are fixed, which remain negotiable, and which technical or organizational assumptions need to be tested before full-scale analysis begins. In this Ten Simple Rules article, we provide a practice-oriented framework for stabilizing inherited computational biology projects before workflows, benchmarks, and decision paths become entrenched. We do not claim that the individual practices described here are novel in isolation. Rather, our contribution is to organize familiar practices into a sequenced framework for a recurrent but under-articulated phase of computational research: inherited-plan execution. Computational biology makes this phase especially important because projects often combine heterogeneous datasets, fragile software environments, undocumented preprocessing choices, benchmarking assumptions, distributed collaborators, and asymmetrical access to contextual knowledge. By making this transition visible and operational, the rules aim to help trainees, supervisors, and collaborators reduce ambiguity, test feasibility, document decisions, and support reproducible and equitable project execution under real-world constraints.

阅读与讨论 → 访问原文 →

探索全球前沿学术脉络