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。

01.
arXiv (CS.CV) 2026-06-19

Single-Stage Hierarchical Rectification for Weakly Supervised Histopathology Segmentation

Existing weakly supervised semantic segmentation (WSSS) methods in computational pathology rely on a multi-stage paradigm: class activation map (CAM) generation, offline pseudo-mask refinement, and fully supervised retraining. While established, this decoupled approach presents fundamental limitations. The multi-stage process not only incurs high computational training costs but also suffers from error propagation: local texture biases in shallow CNN layers generate false-positive artifacts that subsequent refinement steps often fail to correct. To address these persistent challenges through a simple yet highly effective approach, we propose the Single-Stage Hierarchical Rectification (SSHR) framework. Rather than passively refining CAMs post-hoc, our method proactively purifies intermediate feature representations during the forward pass. We introduce a Hierarchical Feature Rectification Module (HFRM) that utilizes deep global semantic context to filter out local anomalies in shallow layers. This mechanism generates high-fidelity activation maps directly within a single training loop. Experiments on the LUAD-HistoSeg and BCSS datasets demonstrate that SSHR outperforms state-of-the-art multi-stage methods. Furthermore, SSHR reduces training duration by 2 to 5 times. This efficiency minimizes computational overhead and accelerates clinical translation for large-scale histopathology workflows. The code is available at: https://github.com/trongduc-nguyen/SSHR

02.
arXiv (quant-ph) 2026-06-16

Towards Interpretability of Neural Quantum States

arXiv:2508.14152v2 Announce Type: replace Abstract: Neural quantum states (NQS) have emerged as a powerful variational ansatz for representing quantum many-body wave functions. Their internal mechanisms, however, remain poorly understood. We investigate the role of correlations for NQS-like quantum state representation by employing a correlation-based interpretable neural network architecture and then proving our observations using Boolean function theory. The correlator neural network demonstrates that, even for simple product states, up to all system-size correlation orders in the chosen computational basis are required to represent a quantum state faithfully. We explain these observations using Fourier expansion, which reveals the correlator basis as the effective basis of the internal NQS structure, the resulting necessity for high-order correlations that is supported by an entanglement bound that scales with the correlation order, consequences of linear dependencies in constrained Hilbert spaces for correlation requirements, and connections between spin basis rotations and the correlator basis. Furthermore, we analyze how neural networks achieve high correlation orders by increasing the magnitude of the network weights, which can be compensated by increasing the network depth. Lastly, we discuss how activation functions, network architectures, and choice of reference basis influence correlation requirements. Our results provide new insights and a better understanding of the internal structure and requirements of NQS, enabling a more systematic use of NQS in future research.

03.
arXiv (CS.CV) 2026-06-15

Gaze Heads: How VLMs Look at What They Describe

How a vision-language model internally solves the task of describing an image is far from obvious. We find that the model develops a specific mechanism for this: a small set of attention heads in its language-model backbone, which we call gaze heads, whose attention tracks the image region the model is currently describing. We find them with a simple correlation score from a few forward passes, using comic strips as a controlled testbed where narrative order is laid out spatially. These gaze heads do not just track the image tokens being described: redirecting their attention to a chosen region forces the VLM to describe that region instead. A single attention-mask intervention on the top-100 gaze heads, fewer than 9% of all heads, steers the model's answer to any chosen comic panel at 83.1% accuracy, while the same intervention on random heads fails to redirect the answer, and intervening on all heads destroys generation. The same lever also extends to continuous control: switching the gaze target mid-generation makes the model wrap up its current panel description and move to the new one within a few tokens. Beyond comics, the same intervention redirects answers to chosen regions in natural COCO images. The mechanism further recurs across model sizes from 2B to 32B parameters and across other VLM architectures, although some frozen-encoder families show no comparable head set. More broadly, this shows that targeted edits identified through mechanistic analysis can serve as practical inference-time levers for steering multimodal model behavior, without any retraining. Our code, interactive demo, and datasets are available at https://gaze.baulab.info/

04.
Nature (Science) 2026-06-10

Light-induced quantum friction of carbon nanotubes in water

Friction slows down moving objects at both macroscopic and microscopic scales1. At the electronic level, quantum friction describes direct transfer of momentum between a liquid and the electrons of a solid2. Owing to its microscopic nature, this phenomenon remains experimentally challenging to capture3. Here we show that near-infrared fluorescent single-walled carbon nanotubes (SWCNTs) exhibit light-induced quantum friction in water. It is measured by observing an excitation-power-dependent linear decrease of around 50% in the diffusion constants of functionalized SWCNTs in aqueous solution. This effect disappears when excitons are localized, as in the case of SWCNTs with quantum defects. We further show that the chemical manipulation of exciton concentration by molecules that increase or decrease SWCNT fluorescence also modulates the diffusion constant by up to a factor of 2. Optical pump terahertz (THz) probe spectroscopy shows an instantaneous response (around 30 cm−1) that we assign to direct exciton–water coupling in the range of water Debye modes. It is followed by an increasing (>100 ps) response in the range of intermolecular translational modes of the hydrogen bond network of water (>100 cm−1), resembling heating. Classical molecular dynamics simulations further support a mechanism in which the fluctuating dipole moments of excitons create frictional forces. These findings establish light-induced quantum friction between excitons in SWCNTs and water and show that electronic excitations can be used to control nanoscale motion and fluid properties. Near-infrared fluorescent carbon nanotubes exhibit light-induced quantum friction in water, in which exciton interactions slow nanoscale motion and enable optical control of diffusion and fluid dynamics.

05.
arXiv (CS.LG) 2026-06-16

Dynamic Link Prediction with Temporally Enhanced Signed Graph Neural Networks

arXiv:2605.26290v2 Announce Type: replace Abstract: Temporal signed networks (TSNs) model the time evolution of cooperative and adversarial relationships that arise in applications such as social media analysis, trust and reputation systems, and financial transaction networks. While graph neural networks (GNNs) perform well for static or unsigned link prediction, effective learning in temporal signed graphs remains challenging due to the interaction of signed relations, evolving structure, and balance-theoretic constraints. To address this gap, we propose a modular temporal enhancement framework for signed GNNs that integrates historical context into otherwise static architectures. The framework introduces a Historical Context Integration Module (HCIM) that combines learnable recency-aware temporal weighting, LSTM-based embedding trajectory modeling, and multi-head temporal attention to capture both short- and long-term signed interaction dynamics. Historical information is fused with current node representations using either global or node-adaptive weighting, allowing the architecture-agnostic framework to accommodate heterogeneous temporal behaviors. We instantiate the approach on the Self-Explainable Signed Graph Transformer (SE-SGformer), preserving interpretability while extending it with temporal awareness. Experiments on real-world and synthetic TSNs, including Bitcoin OTC, Bitcoin Alpha, Reddit, and small-world network models, demonstrate consistent and statistically significant improvements over the static baseline.

06.
arXiv (CS.AI) 2026-06-17

MapSatisfyBench: Benchmarking Satisfaction-Aware Map Agents through Behavior-Grounded Implicit Decision Factors

arXiv:2606.17453v1 Announce Type: new Abstract: Large language model agents are increasingly integrated into map services. Since map services are embedded in everyday-life scenarios rather than professional task settings, users often express their needs informally, resulting in underspecified queries with many unspoken needs, namely, implicit decision factors that are critical for user satisfaction. Although clarification is an effective way to mitigate this issue, it increases user burden in daily interaction, and a capable agent should first proactively recover such factors from available information sources. However, evaluating this ability is challenging. The first challenge is to determine which implicit decision factors are suitable for evaluation. A factor is evaluable only if it affects user acceptance and can be recovered from information available to the agent before it responds. Second, user satisfaction cannot be reliably represented by a single reference answer, requiring a benchmark that converts satisfaction-relevant factors into objective and quantifiable evaluation targets. To address these challenges, we propose a restore-identify-filter framework that reconstructs complete user needs from behavior-chain evidence, identifies implicit decision factors, and retains only those supported by pre-query evidence. Building on this methodology, we construct MapSatisfyBench from large-scale, real-world anonymized user data and annotate ground truth from five dimensions and enables full-chain evaluation of satisfaction-aware map agents. Experiments show that current agents generally perform well on explicit task completion, but remain limited in satisfying implicit decision factors and proactively acquiring the evidence needed for satisfaction-aware decisions. These findings establish MapSatisfyBench as a benchmark for shifting map-agent evaluation from task completion toward satisfaction-aware spatial decision making.

07.
medRxiv (Medicine) 2026-06-15

Dysplasia-Stratified Management of Barrett's Esophagus: An Incidence-Based U.S. Cost-Effectiveness Analysis

作者:

Background and Aims Barrett's esophagus (BE) is the principal precursor of esophageal adenocarcinoma (EAC), whose incidence has risen sharply in Western countries since the 1960s. Effective, dysplasia stratified surveillance strategies are needed to prevent progression. This study evaluated the cost effectiveness of dysplasia stratified surveillance intervals and endoscopic eradication therapy (EET) across the BE spectrum. Methods We developed an incidence-based Markov state transition model of BE progression calibrated to U.S. epidemiologic data from a healthcare sector perspective over a lifetime horizon. Four hypothetical cohorts of 50-year-old individuals with short segment BE (SSBE), nondysplastic BE (NDBE), low grade dysplasia (LGD), or high-grade dysplasia (HGD) were evaluated. Strategies included no surveillance; surveillance at 1-, 2-, 3-, 4-, 5-, or 10-year intervals; standard or AI assisted endoscopy; non endoscopic screening (sponge, breath, miRNA tests); and EET for LGD and HGD. Outcomes included costs, quality adjusted life years (QALYs), incremental cost effectiveness ratios (ICERs), net monetary benefits (NMBs), EAC cases, and EAC-related deaths. Sensitivity analyses used a willingness to pay threshold of US$100,000 per QALY. Results No surveillance was the most cost-effective strategy for SSBE and NDBE. For LGD, upfront EET was more cost effective than all surveillance strategies, with results sensitive to EAC incidence and recurrence. For HGD, EET was cost saving and yielded the greatest QALYs, with findings robust in 99.9% of simulations. EET prevented 12,614 and 44,295 EAC related deaths per 100,000 individuals with LGD and HGD, respectively. Conclusion Dysplasia-stratified management is essential for optimizing surveillance and treatment strategies in BE. Any degree of dysplasia should receive EET followed by targeted post-treatment monitoring, establishing EET as the central therapeutic pathway for dysplastic BE.

09.
arXiv (CS.LG) 2026-06-17

Price of metric universality in vector quantization is at most 0.11 bit

arXiv:2602.05790v2 Announce Type: replace-cross Abstract: Fast computation of a matrix product $W^\top X$ is a workhorse of modern LLMs. To make their deployment more efficient, a popular approach is that of using a low-precision approximation $\widehat W$ in place of true $W$ (``weight-only quantization''). Information theory demonstrates that an optimal algorithm for reducing precision of $W$ depends on the (second order) statistics of $X$ and requires a careful alignment of vector quantization codebook with PCA directions of $X$ (a process known as ``waterfilling allocation''). Dependence of the codebook on statistics of $X$, however, is highly impractical. This paper proves that there exist a universal codebook that is simultaneously near-optimal for all possible statistics of $X$, in the sense of being at least as good as an $X$-adapted waterfilling codebook with rate reduced by 0.11 bit per dimension in the case when $W$ is Gaussian. Such universal codebook would be an ideal candidate for the low-precision storage format, a topic of active modern research, but alas the existence proof is non-constructive. Equivalently, our result shows existence of a net in $\mathbb{R}^n$ that is a nearly-optimal covering of a sphere simultaneously with respect to all Hilbert norms.

10.
medRxiv (Medicine) 2026-06-15

Comparative Analysis of Machine Learning Models vs. Traditional Clinical Calculators for Cardiovascular Risk Prediction

Background: Cardiovascular diseases (CVD) remain the leading global cause of mortality, responsible for approximately 31% of all deaths worldwide in 2021. Traditional risk calculators, including Framingham, ASCVD, SCORE, and SCORE2, have long constituted the cornerstone of primary prevention strategies; however, they were derived predominantly from high-income European and North American populations, thereby limiting their predictive accuracy in diverse epidemiological contexts, particularly among Hispanic/Latino communities. Machine learning (ML) offers an alternative to capture the non-linear interactions inherent in biomedical data. Objective: The present study develops and validates ML-based models for cardiovascular mortality prediction using the National Health and Nutrition Examination Survey (NHANES) 1999-2018 dataset, and systematically compares their discriminative performance against eleven conventional clinical CVD risk calculators. Materials and Methods: A dedicated software platform, "CardioPrediQ," was designed to integrate multiple CVD calculators with ML-based risk assessment. A cohort of 12,847 participants with 16 predictor variables was derived from NHANES. Six algorithms (Logistic Regression, Cox Proportional Hazards, Gradient Boosting, AdaBoost, Random Forest, and Extra Trees) were trained in combination with six class-balancing strategies, yielding 36 model configurations. All models were trained on a stratified 70/30 split and calibrated using the Saerens prior probability adjustment method. Performance was evaluated using AUC-ROC, sensitivity, specificity, F1-score, and a weighted composite score. DeLong's test was employed to assess the statistical significance of AUC differences between the best-performing ML model and each conventional calculator. Results: Gradient Boosting with 2:1 oversampling and Saerens calibration achieved the best overall performance (AUC = 0.8934; composite score = 0.7904), outperforming all traditional calculators in composite ranking. The top six positions were occupied exclusively by ML and statistical models. The mean age of cardiovascular decedents was 67.43 years compared with 47.74 years among survivors. DeLong's test confirmed statistical superiority over six traditional CVD calculators (p < 0.05), whereas the difference against the top-performing calculators (ASCVD, HEARTS Caribbean, ASCVD Colombia, SCORE2, HEARTS North America) did not reach statistical significance. Age dominated feature importance at 41.2% relative weight, followed by systolic blood pressure (18.7%). Saerens calibration reduced the Brier score from 0.1286 to 0.1158, substantially improving probability calibration. Conclusions: ML models demonstrated superior composite performance over traditional calculators. The statistical equivalence with the highest-performing conventional calculators in the NHANES cohort is context-dependent and validates the methodological pipeline. The CardioPrediQ platform addresses the critical need for integrated, scalable CVD risk assessment tools, which is particularly relevant for Latin American populations where calculator validation remains limited. These findings support the integration of calibrated ML-based risk prediction into clinical practice while underscoring the importance of probability calibration for informed clinical decision-making.

11.
arXiv (CS.AI) 2026-06-17

LLMCodec: Adapting Video Codecs for Efficient Weight Compression of Large Language Models

arXiv:2606.05861v2 Announce Type: replace-cross Abstract: The rapid development of large language models(LLMs) has led to remarkable advances in natural language processing. However, the increasing scale of these models introduces substantial challenges in terms of storage, transmission, and deployment. Though great efforts have been devoted to model compression and quantization, existing methods often rely on fine-tuning or calibration data, which exhibit limited generalization across different tensor types. In this paper, we argue that video codecs offer a promising solution for LLM compression, due to their inherent compatibility with matrix structured data, configurable compression strategies, and the availability of highly optimized, off-the-shelf implementations. Therefore, we present LLMCodec, a video codec-based LLM compression method that integrates affine quantization with the recent VVC/H.266 video codec. Beyond VVC, we further compare a range of video codecs and encoding profiles to evaluate their impact on compression performance. Experiments on different models demonstrate the robustness and generality of LLMCodec. Notably, on LLaMA-3-8B at 2-bit precision, LLMCodec reduces perplexity by over 1.5x and improves downstream task accuracy by 21% compared with the existing method.

12.
arXiv (CS.CV) 2026-06-16

ATV-Net: Adaptive Triple-View Network with Dynamic Feature Fusion

Recent advances in semantic segmentation rely heavily on attention-based and transformer-style architectures that, while accurate, introduce considerable architectural complexity and computational cost. This paper asks whether a compact CNN-based segmentation head can remain competitive by adaptively selecting useful receptive-field evidence. We propose ATV-Net, an Adaptive Triple-View Network that attaches a lightweight head to a conventional backbone. The head organizes three complementary views – point-wise, neighborhood-level, and enlarged context – and fuses them through an Adaptive Decision Gate that generates image-dependent weights from global feature statistics. This allows the model to emphasize different receptive-field responses according to scene content, without dense attention or multi-scale aggregation. Experiments on Cityscapes and Pascal VOC 2012 show that ATV-Net achieves 80.31% mIoU on Cityscapes with ResNet-101 and 80.90% with ConvNeXt-Tiny, and 86.7% and 88.5% mIoU on Pascal VOC 2012, respectively, while requiring fewer GFLOPs than representative context-aggregation and attention-based heads. The results indicate that adaptive receptive-field selection remains a practical and effective design choice for CNN-based semantic segmentation.

13.
arXiv (CS.LG) 2026-06-19

Capturing Intransitive Dominance in Tennis Forecasting: A Graph Neural Network Approach

arXiv:2510.20454v2 Announce Type: replace Abstract: Intransitive player dominance, where player A beats B, B beats C, but C beats A, is common in competitive tennis. Yet, there are few known attempts to incorporate it within forecasting methods. We address this problem with a graph neural network approach that explicitly models these intransitive relationships through temporal directed graphs, with players as nodes and their historical match outcomes as directed edges. Our model (65.7% accuracy, 0.214 Brier score) forecasts competitively with established rating systems such as Weighted Elo. Although it does not improve on the baseline in unconditional accuracy, a forecast-encompassing test shows that it carries complementary information. A combined forecast significantly outperforms Weighted Elo, and there is some indication that the gain grows more strongly on the intransitive matchups our model targets. A graph-based representation of player interactions thus captures a forecasting signal that transitive rating systems discard, even between players who share no common opponents.

15.
arXiv (CS.AI) 2026-06-16

Frontier: Towards Comprehensive and Accurate LLM Inference Simulation

arXiv:2605.21312v2 Announce Type: replace-cross Abstract: Modern LLM serving is no longer homogeneous or monolithic. Production systems now combine disaggregated execution, complex parallelism, runtime optimizations, and stateful workloads such as reasoning, agents, and RL rollouts. Simulation is attractive for exploring this growing design space, yet existing simulators lack the architectural completeness and decision-grade fidelity it demands. Their monolithic-replica abstractions are ill-suited to disaggregated serving, while average-case analytical proxies can distort SLA predictions and even reverse optimization conclusions. We present Frontier, a discrete-event simulator for modern LLM inference serving. Frontier features a disaggregated abstraction. It captures the structure and dynamics of modern serving systems by modeling co-location, Prefill-Decode Disaggregation (PDD), and Attention-FFN Disaggregation (AFD) with role-specific cluster workers, incorporating key runtime optimizations (e.g., CUDA Graphs, speculative decoding) within the scheduler-batch-engine loop, and supporting stateful requests for emerging workloads. It further provides accurate and generalizable predictions of computation, communication, and memory costs across diverse serving scenarios with complex workload compositions. On 16-H800 GPU testbed, Frontier achieves an average throughput error below 4%. Compared with state-of-the-art simulators, it reduces end-to-end latency error from 44.9% to 6.4% under co-location and from 51.7% to 2.6% under disaggregation. It scales to over 1K GPUs on commodity CPUs and enables new use cases such as SLA-dependent Pareto frontier exploration, heterogeneous disaggregated allocation, agentic reasoning scheduling validation, and RL post-training reconfiguration. We release Frontier at https://github.com/NetX-lab/Frontier.

16.
arXiv (CS.CL) 2026-06-12

Small LLMs for Biomedical Claim Verification: Cost-Effective Fine-Tuning, Structural Dataset Shortcuts, and Cross-Domain Generalization

作者:

Large Language Models such as GPT-4o and GPT-5 achieve strong zero-shot performance on biomedical claim verification, but cost and opacity limit scalable use. We fine-tune three small LLMs: Phi-3-mini (3.8B), Qwen2.5-3B, and Mistral-7B, via QLoRA on SciFact and HealthVer, providing the first study of QLoRA models against GPT-4o and fine-tuned BioLinkBERT encoders. Mistral-7B QLoRA surpasses both GPT-4o and GPT-5 (up to 12% F1 gain) at a fractional cost using just 1,008 training examples. We conduct extensive in-domain and cross-domain evaluation: models trained on SciFact tested on HealthVer and vice versa, at matched sizes to isolate dataset structure from data quantity. We identify a previously unreported structural artifact in SciFact that inflates in-domain scores, and show through bidirectional out-of-domain evaluation that training on structurally sound data enables robust cross-domain transfer. We plan to release all code and adapter checkpoints.

17.
arXiv (CS.CV) 2026-06-16

Intrinsic Gradient Suppression for Label-Noise Prompt Tuning in Vision-Language Models

Contrastive vision-language models like CLIP exhibit remarkable zero-shot generalization. However, prompt tuning remains highly sensitive to label noise, as mislabeled samples generate disproportionately large gradients that can overwhelm pre-trained priors. We argue that because CLIP already provides a near-optimal initialization, adaptation should be inherently conservative, particularly against the extreme gradient updates common in noisy settings. To this end, we propose Double-Softmax Prompt Tuning (DSPT), a hyperparameter-free method for intrinsic gradient suppression. By applying a sequential probabilistic normalization, DSPT induces a self-adaptive saturation zone that suppresses gradients from high-error noisy samples while maintaining informative updates. We also provide both theoretical analysis and empirical evidence about how this mechanism achieves adaptive suppression. This design transforms ``gradient vanishing'', traditionally a training bottleneck, into a principled noise-filtering shield for label-noise prompt tuning. Extensive experiments confirm that this simple, drop-in design achieves state-of-the-art robustness across various noisy benchmarks, outperforming methods with complex architectures and handcrafted hyperparameters.

18.
arXiv (CS.LG) 2026-06-15

On the Generalization Bounds of Symbolic Regression with Genetic Programming

arXiv:2604.17402v2 Announce Type: replace Abstract: Symbolic regression (SR) with genetic programming (GP) aims to discover interpretable mathematical expressions directly from data. Despite its strong empirical success, the theoretical understanding of why GP-based SR generalizes beyond the training data remains limited. In this work, we provide a learning-theoretic analysis of SR models represented as expression trees. We derive a generalization bound for GP-style SR under constraints on tree size, depth, and learnable constants. Our result decomposes the generalization gap into two interpretable components: a structure-selection term, reflecting the combinatorial complexity of choosing an expression-tree structure, and a constant-fitting term, capturing the complexity of optimizing numerical constants within a fixed structure. This decomposition provides a theoretical perspective on several widely used practices in GP, including parsimony pressure, depth limits, numerically stable operators, and interval arithmetic. In particular, our analysis shows how structural restrictions reduce hypothesis-class growth while stability mechanisms control the sensitivity of predictions to parameter perturbations. By linking these practical design choices to explicit complexity terms in the generalization bound, our work offers a principled explanation for commonly observed empirical behaviors in GP-based SR and contributes towards a more rigorous understanding of its generalization properties.

19.
arXiv (CS.CV) 2026-06-17

Does the Data Processing Inequality Reflect Practice? On the Utility of Low-Level Tasks

The data processing inequality is an information-theoretic principle stating that the information content of a signal cannot be increased by processing the observations. In particular, it suggests that there is no benefit in enhancing the signal or encoding it before addressing a classification problem. This assertion can be proven to be true for the case of the optimal Bayes classifier. However, in practice, it is common to perform "low-level" tasks before "high-level" downstream tasks despite the overwhelming capabilities of modern deep neural networks. In this paper, we aim to understand when and why low-level processing can be beneficial for classification. We present a comprehensive theoretical study of a binary classification setup, where we consider a classifier that is tightly connected to the optimal Bayes classifier and converges to it as the number of training samples increases. We prove that for any finite number of training samples, there exists a pre-classification processing that improves the classification accuracy. We also explore the effect of class separation, training set size, and class balance on the relative gain from this procedure. We support our theory with an empirical investigation of the theoretical setup. Finally, we conduct an empirical study where we investigate the effect of denoising and encoding on the performance of practical deep classifiers on benchmark datasets. Specifically, we vary the size and class distribution of the training set, and the noise level, and demonstrate trends that are consistent with our theoretical results.

20.
arXiv (CS.LG) 2026-06-16

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

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.

21.
arXiv (CS.AI) 2026-06-16

SkillVetBench: LLM-as-Judge for Multi-Dimensional Security Risk Evaluation in Open-Source LLM Agent Skills

arXiv:2606.15899v1 Announce Type: cross Abstract: Open-source LLM agent ecosystems are growing rapidly, yet the security of community-contributed skills - modular tool definitions that extend agent capabilities - remains largely unvetted. The gap we fill: existing scanners operate at the code layer and are structurally blind to instruction-layer and multi-agent risk - natural-language directives that hijack an agent, exfiltrate data through encoded side channels, or chain harm across pipelines - so what is needed is a semantic, multi-dimensional vetting system rather than another signature matcher. We present SKILLVETBENCH, a live public leaderboard on Hugging Face that uses an LLM-as-Judge to vet agent skills. What is new: SARS (Skill Agentic Risk Score), a five-dimensional agentic-risk metric with a principled weighted formula for instruction-following systems. What is integrated: full CVSS v4.0 vector decomposition and a ClawHub dual-view that places our LLM-generated review beside the official marketplace verdict. What is demonstrated: drawing on our companion benchmark paper [ 1], the LLM-as-Judge stage achieves zero false negatives across 78 confirmed-malicious skills and zero false positives across 22 benign controls, while the best static baseline (SKILLSIEVE) still misses 15%; for instruction-layer categories such as Prompt Injection and Memory Poisoning, conventional tools miss between 89% and 100% of threats (e.g., CODEBERT detects none of nine memory-poisoning skills). Detection rates vary from 35% to 95% across four LLM evaluators, motivating ensemble scoring in production deployments.

22.
arXiv (CS.LG) 2026-06-17

Evaluating Open-Source LLMs for Multi-Label ATT&CK Technique Classification on CTI Reports

arXiv:2606.18166v1 Announce Type: cross Abstract: Classifying Cyber Threat Intelligence (CTI) using MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) is essential for proactive defense, but historically required extensive human effort. Pre-Large Language Model (LLM) automation sped up this process, but could not resolve the complex language and multi-step attack patterns found in unstructured CTI reports. LLMs addressed previous limitations by using contextual reasoning to understand unstructured text. However, current evaluations rely on simplified, single-technique sentences that ignore the complexity of real-world CTI reports, which often leads to inflated performance results. Consequently, the baseline performance of open-source LLMs on complex unstructured CTI reports remains unevaluated. To address this gap, we constructed a ground-truth dataset of 2,076 human-annotated sentences (1,281 technique-positive, 795 negative) from 83 complex unstructured CTI reports. These sentences were mapped to 114 unique ATT&CK techniques using a six-phase annotation process, achieving \k{appa} = 0.68 inter-annotator agreement. Using this dataset, we evaluated seven open-source LLMs ranging from 8B to 236B parameters across prompt strategy and temperature configurations. The highest-performing LLM achieved a micro-averaged F1 score of 0.22, establishing the empirical baseline for multi-label ATT&CK classification on complex unstructured CTI. Parameter size showed a statistically significant positive correlation with F1 score. Prompt strategy and temperature produced no statistically significant gains across model configurations. These results indicate that current open-source LLMs are insufficient for production-grade ATT&CK classification. The dataset, benchmark, and findings provide a reproducible foundation for future CTI research.

23.
arXiv (CS.AI) 2026-06-15

The Weight Norm Sets the Grokking Timescale: A Causal Delay Law

arXiv:2606.13753v1 Announce Type: cross Abstract: Grokking is the delayed onset of generalization in neural networks, arising long after they fit the training data. Whether the weight norm causes this delay is disputed: some studies report a critical norm at the transition, others observe grokking with no fixed norm at all. We settle this by intervening on the norm during training rather than only observing it. Under free training with weight decay, networks grok when the weight norm reaches a value Wc that varies little across seeds and learning rates (CV 1 to 2 percent) and grows with the modular base as a power law. When we instead clamp the norm to a fixed multiple rho of Wc and hold it there, the network still groks, but the delay follows T_grok proportional to exp(alpha rho). One exponent, alpha near 7.5, fits this delay across four moduli (R^2 = 0.996). Over the swept ranges the held norm moves the delay by about 19x and the learning rate by only about 2x, and holding the norm above Wc slows grokking rather than preventing it. A final LayerNorm removes the dependence by decoupling weight scale from the network function; without it the exponential law returns. This pinned-norm delay is the exponential counterpart to the logarithmic delay predicted for a freely contracting norm.

24.
arXiv (CS.CL) 2026-06-11

Grammar-Constrained Decoding Can Jailbreak LLMs into Generating Malicious Code

Large Language Models (LLMs) are increasingly used for code generation, raising concerns that they may be misused to produce malicious code. Meanwhile, Grammar-Constrained Decoding (GCD) has been widely adopted to improve the reliability of LLM-generated code by enforcing syntactic validity. In this paper, we reveal a counterintuitive risk: this reliability-oriented technique can itself become an attack surface. We uncover a new jailbreak attack, termed CodeSpear, that exploits GCD to induce LLMs into generating malicious code. Our experiments show that simply applying a benign code grammar constraint can effectively jailbreak LLMs. To address this vulnerability, we propose CodeShield, a safety alignment approach that robustly preserves safe behavior even under attacker-controlled grammar constraints. CodeShield aligns the model in the code modality by teaching it to generate honeypot code under GCD. Such code is semantically harmless, so it does not implement the malicious request, and structurally diverse, so it is difficult to suppress through grammar tightening. At the same time, CodeShield still preserves natural-language refusals when natural language is available. Experiments on 10 popular LLMs across 4 benchmarks show that CodeSpear outperforms representative jailbreak baselines and increases the attack success rate by more than 30 percentage points on average. CodeShield also restores safety under CodeSpear while preserving benign utility. Our findings reveal a fundamental risk of GCD and call for greater attention to its potential security implications.

25.
arXiv (CS.LG) 2026-06-18

Hierarchical Planning with Latent World Models

arXiv:2604.03208v2 Announce Type: replace Abstract: World models are a promising path to zero-shot embodied control through planning. However, existing world model planners struggle on long-horizon, multi-stage tasks: prediction errors compound and naive search is exponential in the planning horizon. Hierarchy mitigates both by decomposing tasks into shorter, tractable subproblems; yet prior hierarchical approaches either amortize control into task-specific policies (hierarchical RL) or assume low-dimensional states and known dynamics (classical hierarchical MPC). We present Hierarchical Planning with Latent World Models (HWM), an architecture and planning paradigm for hierarchical model predictive control (MPC) directly on visual world models trained solely via next-latent prediction. HWM learns world models at multiple temporal scales within a shared latent space, so predictions from the long-horizon model serve as subgoals for the short-horizon model via latent matching, without task-specific rewards, skill learning, or hierarchical policies. To keep long-horizon search tractable, HWM learns an action encoder that compresses primitive action chunks into latent macro-actions. On real-world Franka manipulation, HWM solves pick-and-place from a single goal image at 70% success vs. 0% for single-level planning. Across simulated push manipulation and maze navigation, HWM consistently improves performance on long-horizon tasks while requiring up to 3x less planning compute.