跳到正文
研踪PaperTrail

YOUR RESEARCH, IN SIGHT

让下一篇重要论文，不再错过。

覆盖计算机全学科，以 AI 为快捷入口。关注方向、作者和关键词，把发现变成持续的研究积累。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+2471新增论文
+2025其中 AI 相关
+163已有论文修订
+100新增 AI 导读

arXiv 最近完成扫描：2026-11-10 18:29 · 全站最近入库：2026-11-10 15:22各信源进度 →

统计口径与数据新鲜度

统计 2026-11-09 18:54 — 2026-11-10 18:54（北京时间） · 点击数字查看对应简报；AI 相关包含在新增论文内。

最近完成检查：2026-11-10 18:29最近发现新增版本：2026-11-10 15:22库内论文最新更新时间：2026-11-10 02:59

每分钟刷新已保存的数据。空检查不代表采集异常；新收录可能包含历史论文。AI 导读按新增版本计数。

20,643 已收录论文17,531 AI 相关5,643 有修订112 HF 社区发现

浏览论文库

研究日报 / 周报 →按信源浏览 →

时间范围跟随排序：版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常，查看状态 →

找到 20,643 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-10 18:28（北京时间）
版本更新 · v2arXiv:2609.22299

When Does Test-Time Physical Diagnosis Pay?

Zhengshu Zhang

原文摘要

When a robot faces unfamiliar physical conditions, a common approach is to collect evidence about what changed and adapt. For such diagnosis to improve behavior, six ordered empirical conditions must hold: a meaningful reference, identifiability of the physical condition, use of the acquired evidence, decision value, selection value over a fixed alternative, and safe realization. We test this chain in controlled and public environments. In our controlled environments the frozen policy chooses correctly on every decision, but which evidence it buys is fixed by how the panels assign decisions to evidence strata. On three mechanisms absent from training, the frozen decoder fails its registered transfer gate. On all 96 decisions of the transfer panel, its full-trace input lies far outside its training range, and on such inputs it returns one answer whatever the trace contains. All 15 of its errors fall in the 30 decisions from which cheaper evidence was removed by design, a split hidden by aggregate accuracy. None of three primary public tasks passes its registered qualification gates. Before attributing a failure to how a frozen model uses evidence, evaluation should therefore check whether its deployment inputs lie inside its training range.

cs.RO · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2601.01535

Improving Flexible Image Tokenizers for Autoregressive Image Generation

Zixuan Fu, Lanqing Guo, Chong Wang, Binbin Song, Ding Liu, Bihan Wen

原文摘要

Flexible image tokenizers aim to represent an image using an ordered 1D variable-length token sequence. This flexible tokenization is typically achieved through nested dropout, where a portion of trailing tokens is randomly truncated during training, and the image is reconstructed using the remaining preceding sequence. However, this tail-truncation strategy inherently concentrates the image information in the early tokens, limiting the effectiveness of downstream AutoRegressive (AR) image generation as the token length increases. To overcome these limitations, we propose \textbf{ReToK}, a flexible tokenizer with \underline{Re}dundant \underline{Tok}en Padding and Hierarchical Semantic Regularization, designed to fully exploit all tokens for enhanced latent modeling. Specifically, we introduce \textbf{Redundant Token Padding} to activate tail tokens more frequently, thereby alleviating information over-concentration in the early tokens. In addition, we apply \textbf{Hierarchical Semantic Regularization} to align the decoding features of earlier tokens with those from a pre-trained vision foundation model, while progressively reducing the regularization strength toward the tail to allow finer low-level detail reconstruction. Extensive experiments demonstrate the effectiveness of ReTok: on ImageNet 256$\times$256, our method achieves superior generation performance compared with both flexible and fixed-length tokenizers. Code will be available at: \href{https://github.com/zfu006/ReTok}{https://github.com/zfu006/ReTok}

cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2606.25147

TokenMinds: Pretrained User Tokens and Embeddings for User Understanding in Large Recommender Systems

Qingyun Liu, Bo Yan, Yang Liu, Yuji Roh, Ekansh Sharma, Likang Yin, Emma Olowo, Yuxuan Li 等 19 位作者

原文摘要

User modeling in industrial recommender systems typically produces dense embeddings, which suffer from representational constraints due to fixed-dimensional vectors. As an emerging alternative for discrete user representation, LLM-generated natural-language user profiles often capture language semantics rather than sequential patterns in user behaviors. Meanwhile, Semantic ID (SID) is effective for item representation, yet existing discrete user representations remain largely confined to quantizing already-pooled user vectors, clustering historical watches, or predicting coarse topic clusters solely for candidate retrieval. We propose TokenMinds, an industrial system that extends the PLUM framework from SID-based item retrieval to generative sequential user modeling. Built on an encoder-decoder architecture adapted from pretrained LLMs, TokenMinds introduces a dual-output design: the encoder outputs a dense user embedding capturing a holistic user state, while the decoder generates discrete SID-based user tokens representing multi-faceted future interests. Grounded in item semantics, these tokens preserve distinct interests that a single pooled vector would average away. Additionally, a shared SID vocabulary across long- and short-form videos unifies cross-scenario behaviors into a single model, cutting upstream training and serving compute by 50% and 31%. Furthermore, leveraging the text vocabulary inherited from the pretrained LLM, TokenMinds interleaves textual search queries directly with watch histories to capture explicit user intent. We validate TokenMinds through offline evaluations and live A/B experiments across YouTube's primary ranking surfaces, serving billions of users asynchronously: SID-based user tokens are effective on their own, yield additive gains alongside dense embeddings, and together outperform the previous-generation, non-LLM user model in production.

cs.IR · cs.AI · cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36379

Evaluating Federated Vision-Language and Vision-Only Models for Privacy-Aware Breast Cancer Imaging

Malithi Mithsara Wanniarachchi Kankanamge, Ning Yang, Sree Bash Debnath, Md Zobaer Islam, Abdur R. Shahid

原文摘要

Artificial intelligence (AI) shows promise for breast imaging, but limited labeled data, institutional data fragmentation, and privacy concerns hinder centralized training. We investigate a privacy-aware federated learning framework combining pretrained vision-language models (VLMs) with client-level differential privacy. We compare general-purpose CLIP models, biomedical BiomedCLIP, and an ImageNet-pretrained ViT-B/16 baseline on two public breast imaging datasets, with and without differential privacy. We also evaluate a CLIP ViT-B/16 ablation that removes the text pathway. Evaluation includes patient-level ROC-AUC with bootstrap 95% confidence intervals, baseline comparisons, threshold sensitivity, inter-model agreement, and communication cost. Under federated learning, all five models achieve high normal-versus-sick discrimination (ROC-AUC: 0.940-0.990). However, acquisition year alone achieves 0.974 on the same test partition, raising concerns about confounding rather than genuine pathology recognition. Benign-versus-malignant discrimination remains uncertain (ROC-AUC: 0.504-0.637), with all confidence intervals including 0.50. Removing the text pathway preserves classification accuracy while halving communication cost. Differential privacy reduces ROC-AUC by 0.375, attributable to perturbing the full model state rather than only trainable parameters. These findings highlight the importance of evaluating confounding, predictive reliability, privacy implementation, and deployment efficiency when developing federated AI systems for breast imaging.

cs.CV · cs.CR · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v4arXiv:2601.21092

MapPFN: Learning Causal Perturbation Maps in Context

Marvin Sextro, Weronika Kłos, Gabriel Dernbach

原文摘要

Planning effective interventions in biological systems requires treatment-effect models that adapt to unseen biological contexts by identifying their specific underlying mechanisms. Yet single-cell perturbation datasets span only a handful of biological contexts, and existing methods cannot leverage new interventional evidence at inference time to adapt beyond their training data. To meta-learn a perturbation effect estimator, we present MapPFN, a prior-data fitted network (PFN) pre-trained on a synthetic biological prior with causal interventions, decoupling pre-training from limited wet-lab data. Unlike existing methods, MapPFN uses in-context learning to map a sequence of experiments to a post-perturbation distribution, enabling a single pre-trained model to adapt to new datasets and gene sets. Our results show that combining synthetic pre-training with fine-tuning on real single-cell data improves perturbation predictions across biological contexts. The code, model and data are available at https://marvinsxtr.github.io/MapPFN.

cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2606.12232

Re-evaluating Confidence Remasking in Masked Diffusion Language Models

Stipe Frković, Metod Jazbec, Dan Zhang, Christian A. Naesseth, Ilija Bogunovic, Eric Nalisnick

原文摘要

Masked diffusion language models (dLLMs) have recently emerged as a competitive alternative to autoregressive language models, with the promise of faster inference via parallel token generation. A notable limitation of the masked formulation, however, is that once a token has been unmasked it can no longer be revised, leaving dLLMs vulnerable to early sampling mistakes. Confidence-based remasking methods aim to address this in a training-free manner by revising low-confidence tokens, with encouraging early results. In this work, we compare the accuracy-compute Pareto frontiers of three such methods (WINO, Saber, and COVER) to those of confidence-based unmasking alone (Fast-dLLM). We find that under greedy decoding, all three perform on par with Fast-dLLM at commonly used shorter block lengths; gains of remasking appear only at longer block lengths, where confidence-based unmasking is known to degrade. Under stochastic decoding, WINO reduces generation diversity beyond that of unmasking alone, while it helps recover from errors introduced by learned Bernoulli unmasking policies. Our results show that the benefits of confidence-based remasking are highly setting-dependent, underscoring the need for more comprehensive evaluation. Finally, our analyses point to the dLLM, not the remasking rule, as the bottleneck, suggesting that effective remasking may require changes to training, not only to inference.

cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36374

HIDRA: Hierarchical Dual-Routing Attention for Replay-Free Lifelong Imitation Learning

Fanqi Yu, Matteo Tiezzi, Cigdem Beyan, Tommaso Apicella, Vittorio Murino

原文摘要

Robotic agents operating in real-world environments must adapt continuously from a stream of multimodal demonstrations, often under constraints that preclude storing or revisiting past data. In this replay-free, single-pass lifelong imitation learning setting, sequential updates can progressively distort the representation space, making expert-based approaches particularly sensitive to routing interference, causing expert selection to degrade over time. We propose HIDRA, a structured routing framework that mitigates this issue through a hierarchical dual-attention mechanism, which decouples instruction-level expert retrieval from context-dependent refinement. To further stabilize routing under representation drift, HIDRA introduces a key-level geometric regularization that enforces both alignment within tasks and separation across tasks. The resulting approach enables reliable expert selection while supporting an expandable set of experts for continual adaptation. We evaluate HIDRA on several LIBERO benchmark suites, including more challenging variants with heterogeneous task sequences and paraphrased language instructions. Our approach consistently outperforms both replay-free and replay-based baselines, improving AUC and forward transfer while maintaining low forgetting, with stronger gains under high interference and language variation. The code is available at: https://github.com/yfqi/HIDRA.

cs.RO · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.17709

Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning

Yusuf Umut Ciftci, Joonwon Kang, Ajay Suresha Sathya, Somil Bansal

原文摘要

Imitation learning typically treats every demonstration state as equally important, optimizing a behavior-cloning objective uniformly across the dataset. Yet not all states are created equal: in a small number of critical states, even a slight deviation from the expert action can lead to a failure, while in many other states such errors are recoverable and inconsequential. We formalize state criticality in imitation learning as the worst-case degradation in downstream value induced by a bounded perturbation of the expert action at a given state. We then propose SCRIM, a practical State CRIticality Measure that estimates this quantity by training an adversarial disturbance policy and a value critic around a fixed imitation policy. SCRIM assigns each demonstration state a score reflecting how vulnerable the task outcome is to small, policy-relevant action errors. We show that these scores can be used to reweight behavior-cloning losses, improving imitation at consequential states and increasing overall task performance. We further show that SCRIM enables data curation by selecting compact, high-value subsets of demonstrations that preserve performance while substantially reducing the amount of training data. We evaluate SCRIM across various simulated and real-world robotic manipulation tasks, and demonstrate that SCRIM improves policy training, enables efficient data selection, and provides interpretable insights into which moments in a demonstration are most critical for task success. Project website: https://sia-lab-git.github.io/SCRIM/

cs.LG · cs.AI · cs.RO

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.23857

Strong Teacher Not Needed? On Distillation in LLM Pretraining

Taiming Lu, Zhuang Liu

原文摘要

Knowledge distillation generally assumes a strong-to-weak relationship where stronger teachers yield better students. In this work, we examine this assumption about distillation in large language model pretraining. By varying architecture sizes and training token budgets, we create strong-to-weak, same-level, and weak-to-strong teacher-student relationships, and study distillation's effectiveness under each. We find that the teacher need not be strong: with proper mixing of the language modeling and knowledge distillation losses, even small and undertrained teachers improve larger students. At the same time, a stronger teacher is not always better: pushing the teacher further, through more parameters or more training tokens, can saturate or even reverse the distillation gains. We further observe that distillation improves generalization (out-of-distribution and downstream performance) more readily than in-domain fitting. Together, these results challenge the common belief that distillation pretraining always requires a strong teacher.

cs.LG · cs.CL

查看论文
arXiv ↗PDF ↗
版本更新 · v6arXiv:2407.04495

Speed-accuracy relations for diffusion models: Wisdom from nonequilibrium thermodynamics and optimal transport

Kotaro Ikeda, Tomoya Uda, Daisuke Okanohara, Sosuke Ito

原文摘要

We discuss a connection between a generative model, called the diffusion model, and nonequilibrium thermodynamics for the Fokker-Planck equation, called stochastic thermodynamics. Using techniques from stochastic thermodynamics, we derive the speed-accuracy relations for diffusion models, which are inequalities that relate the accuracy of data generation to the entropy production rate. This relation can be interpreted as the speed of the diffusion dynamics in the absence of the non-conservative force. From a stochastic thermodynamic perspective, our results provide quantitative insight into how best to generate data in diffusion models. The optimal learning protocol is introduced by the geodesic of space of the 2-Wasserstein distance in optimal transport theory. We numerically illustrate the validity of the speed-accuracy relations for diffusion models with different noise schedules and different data. We numerically discuss our results for optimal and suboptimal learning protocols. We also demonstrate the applicability of our results to data generation from the real-world image datasets.

cs.LG · stat.ML

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36371

BoxWorld: Let World Engines Control and Video Generators Render

Zike Wu, Ahmed Hatem, Chintan Trivedi, Wallace Michel Pinto Lira, Xuebin Qin, Lele Wang, Renjie Liao

原文摘要

We tackle the challenge of building real-time interactive video worlds with user-defined behaviors. Such worlds must track object states, apply interaction rules, and render the resulting consequences as users act. Yet existing end-to-end video world models typically learn interaction dynamics within the generator, making it difficult to specify object-level rules and reason about their consequences. In this work, we introduce BoxWorld, which separates interaction reasoning from generative rendering. Given user-specified rules, a coding agent creates a Box World Engine that determines interaction outcomes and exposes the resulting object states through a lightweight interface of layout boxes, persistent object attributes, and dynamic status variables, leaving appearance and physical effects to generative rendering. A Box-State Generative Renderer then renders this evolving state stream into video autoregressively, enabling online control over object motion, state switches, and conditional events. Rendering this state stream is non-trivial: the renderer must respond to object-level state updates in real time while remaining coherent with the generated history. To this end, we introduce a three-stage training strategy that learns object grounding from open-domain videos and state transitions from gameplay videos, both annotated automatically, and finally performs rollout distillation on self-generated histories. We further propose PairedStateBench, a paired benchmark for fine-grained interaction control. BoxWorld substantially outperforms end-to-end world models and remains competitive across multiple public benchmarks. Project page: https://boxworld-web.github.io.

cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2510.06257

Toward Uncertainty-Aware and Generalizable Neural Decoding for Quantum LDPC Codes

Xiangjun Mi, Frank Mueller

原文摘要

Quantum error correction (QEC) is essential for scalable quantum computing, yet decoding errors via conventional algorithms result in limited accuracy (i.e., suppression of logical errors) and high overheads, both of which can be alleviated by inference-based decoders. To date, such machine-learning (ML) decoders lack two key properties crucial for practical fault tolerance: reliable uncertainty quantification and robust generalization to previously unseen QEC codes. To address this gap, we propose a Quantum Bayesian graph Attention decoder \textbf{(QuBA)} that enables expressive error-pattern recognition alongside calibrated uncertainty estimates. Building on QuBA, we further develop a multi-phase training framework with enhanced cross-domain robustness enabling decoding beyond the training set called Sequential Aggregate Generalization under Uncertainty \textbf{(SAGU)}. Experiments on bivariate bicycle (BB) codes and their coprime variants demonstrate that (i)QuBA consistently outperforms the classical baseline belief propagation (BP), achieving up to a \emph{two orders of magnitude} reduction in logical error rate (LER) under confident-decision bounds on the coprime BB code $[[154,6,16]]$; (ii) SAGU achieves decoding performance comparable to or even outperforming QuBA's domain-specific training approach.

quant-ph · cs.IT · cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36370

IOSage: Detecting HPC I/O Bottleneck Patterns to Guide LLM Recommendations

Mahdi Banisharif, Tabassum Mahmud, Ahmad Tarraf, Suren Byna, Ali Jannesari

原文摘要

HPC applications often leave much of their parallel file system's bandwidth unused, and turning their Darshan logs into concrete I/O fixes still requires a scarce I/O expert. Rule-based tools flag counters against thresholds set by experts, and in tools built on large language models (LLMs), such as ION and IOAgent, the LLM decides which bottlenecks a log contains. We present IOSage, in which a classifier trained on many labeled logs, not the LLM, makes that decision, since a single log gives an LLM no baseline for what is abnormal. The LLM receives the I/O bottleneck patterns the classifier detects, the job's configuration files, per-file Darshan evidence, and for each pattern a card built from benchmark training runs, and writes recommended changes. Applied by hand and measured in repeated paired runs, the changes IOSage names speed up median total wall time 4.38x for an openPMD kernel (4.02x on a second storage tier) and 1.42x for hourly WRF restarts. For 5 of the 6 application scenarios with a helpful change, IOSage with Claude Sonnet 4 names it, including the change for 10-minute WRF restarts (up to 1.93x faster); with GPT-4o and Llama 3.1 as well, the classifier and cards name more helpful changes than the LLM alone. On 78 benchmark test runs without a pattern, the LLM deciding alone proposes changes for 64; with the classifier, for 9. The classifier reaches 0.929 Micro-F1 on 488 labeled benchmark test runs (0.916 as the gate; 0.832 on the 158 from configurations and jobs outside training), above Drishti's rules, WisIO, and IOAgent's pipeline without its literature retrieval. On TraceBench's nine expert-labeled traces from production systems, it has the highest precision of the tools compared (0.82), with 2 false positives against 12 for the keyword-mapped LLM diagnoses shipped with TraceBench and 5 for Drishti, on the five issue labels that match registered patterns.

cs.DC · cs.LG · cs.PF

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36367

MUSE: Revealing Responsiveness Gaps of LLM Tutors on Multiple-Solution Programming Problems

Sky Cen, Shashank Sonkar

原文摘要

When a problem admits multiple valid solutions, effective tutoring requires building on a student's strategy while introducing alternatives at appropriate points. However, whether LLM tutors can support this flexibility by appropriately responding to students' diverse solution approaches remains largely unexplored. In this work, we introduce MUSE: Multi-solution Understanding and Simulation Environment, a controlled environment for simulating student-tutor conversations on multiple-solution programming problems. Grounded in learning science on responsive instruction and multiple-solution learning, we investigate which solutions LLM tutors prefer, how they respond to students' approaches, and how they transition between solutions. Experiments suggest that LLM tutors exhibit systematic preferences among valid solutions, often favoring brute-force approaches when not providing general advice. More concerningly, tutors often impose these preferences, redirecting students toward their own preferred approach rather than supporting the student's. Yet this tendency reverses once a solution is completed: tutors often remain anchored to the existing approach when a stronger alternative could deepen understanding, and sometimes even transition to a less optimal approach. Thus, tutors struggle both to build on students' reasoning when appropriate and to extend their reasoning when beneficial. Promisingly, targeted post-training improves both tutor responsiveness and transitions toward more optimal solutions.

cs.CY

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.31810

PaperFormalizer: Author-in-the-Loop Autoformalization of Research Papers in Lean

Jiyuan Tan, Jose Blanchet, Vasilis Syrgkanis

原文摘要

Even a carefully proofread paper can leave small gaps in its proofs or its stated assumptions. We introduce PaperFormalizer, a Lean-based agentic pipeline that helps authors find and close such gaps in their own papers and turn each paper into a Lean-verified version. Starting from the LaTeX source, PaperFormalizer transcribes the paper into a dependency graph of statements, audits each statement in natural language for faithfulness, ambiguity and correctness, and then formalizes and proves the statements in Lean against Mathlib, optional domain libraries and the author's personal toolkit. Only the author knows what a statement is meant to say and how a formalization gap in a result is best fixed. The pipeline therefore stops and asks the author whenever a statement admits more than one reading, presents every repair, in natural language and TeX symbols, as a proposal that the author can accept, decline or redirect, and leaves to the author every decision that could potentially change what the paper claims. The output is an annotated and revised version of the paper, a report of the changes, and Lean code in which each declaration is linked to the statement it formalizes. The author's toolkit grows as more papers are formalized, and authors can share their verified papers so that researchers who build on them can reuse their Lean code. By making verified papers practical to produce, PaperFormalizer aims to make research that rests on mathematical proofs more robust. PaperFormalizer runs inside a coding agent (Claude Code or Codex) and is available at https://github.com/Jiyuan-Tan/PaperFormalizer. Papers with Lean code shared by users are available at https://paperformalizer.org.

cs.AI · cs.LO

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36364

When Slides Met Scans: Cross-Attention for Renal Cancer Prognosis

Anusha Krishnamoorthy, Basudha Pal

原文摘要

Histopathology and computed tomography (CT) provide complementary views of renal cancer. We connect these modalities through bidirectional cross-attention between 16 histology prototypes and up to 32 kidney-centered CT tokens from frozen foundation models. In 210 TCGA-KIRC patients with 46 disease-specific survival (DSS) events, 5-fold evaluation yielded a C-index of $0.760\pm0.081$ and 3-year AUROC of 0.773 (95\% CI 0.678--0.860), the highest averages among evaluated unimodal and fusion baselines. External testing in 180 CPTAC-CCRCC patients yielded DSS C-index of 0.629 with CT withheld. Attribution analysis linked predictions to image regions while exposing acquisition artifacts. These findings support further study of cross-scale fusion and independent multimodal validation.

eess.IV · cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.02141

On the Optimal Sample Complexity of Offline Multi-Armed Bandits with KL Regularization

Kaixuan Ji, Qiwei Di, Heyang Zhao, Qingyue Zhao, Quanquan Gu

原文摘要

Kullback-Leibler (KL) regularization is widely used in offline decision-making and offers several benefits, motivating recent work on the sample complexity of offline learning with respect to KL-regularized performance metrics. Nevertheless, the exact sample complexity of KL-regularized offline learning remains largely from fully characterized. In this paper, we study this question in the setting of multi-armed bandits (MABs). We provide a sharp analysis of KL-PCB (Zhao et al., 2026), showing that it achieves a sample complexity of $\tilde{O}(ηSAC^{π^*}/ε)$ under large regularization $η= \tilde{O}(ε^{-1})$, and a sample complexity of $\tildeΩ(SAC^{π^*}/ε^2)$ under small regularization $η= \tildeΩ(ε^{-1})$, where $η$ is the regularization parameter, $S$ is the number of contexts, $A$ is the number of arms, $C^{π^*}$ policy coverage coefficient at the optimal policy $π^*$, $ε$ is the desired sub-optimality, and $\tilde{O}$ and $\tildeΩ$ hide all poly-logarithmic factors. We further provide a pair of sharper sample complexity lower bounds, which matches the upper bounds over the entire range of regularization strengths. Overall, our results provide a nearly complete characterization of offline multi-armed bandits with KL regularization.

cs.LG · cs.AI · math.ST · stat.ML

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36363

Reliability-Aware Short-Term Glycemic Risk Prediction Using an Anchor-Based Comparative Framework of Sequence, Multimodal, and State-Updating Digital-Twin Models

Pir Bakhsh Khokhar, Carmine Gravino, Fabio Palomba

原文摘要

Anticipating short-term hypoglycemic and hyperglycemic risk is essential for proactive diabetes management. However, existing machine learning approaches often use inconsistent evaluation protocols and emphasize predictive performance over reliability. This study introduces an anchor-based benchmarking framework integrating longitudinal glucose sequences and multimodal patient information. Using the T1DiabetesGranada dataset, 5,503,441 standardized prediction anchors were constructed from 736 individuals with type 1 diabetes to evaluate hypoglycemia and hyperglycemia prediction at 30-, 60-, and 120-minute horizons. Three neural architectures, a Hybrid LSTM-Transformer, a Multimodal Digital Twin, and a State-Updating Digital Twin with gap-aware longitudinal state progression, were compared under identical patient-level validation protocols. Beyond predictive performance, the evaluation examined calibration, classification thresholds, explanation reliability, cross-model disagreement, and robustness to controlled perturbations. No statistically significant predictive differences were observed after Holm correction, although the models exhibited complementary reliability characteristics. The Multimodal Digital Twin achieved the lowest mean Brier score, while the State-Updating Digital Twin demonstrated the lowest mean expected calibration error (ECE-10). The Hybrid LSTM-Transformer showed strong explanation faithfulness and stability under its temporal attribution protocol. These findings highlight the importance of evaluating digital-twin models beyond predictive accuracy and emphasize the need for reliability assessment and external validation before clinical deployment.

cs.LG · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2609.14708

Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Composition

Yecheng Wu, Song Han, Han Cai

原文摘要

A core goal of efficient reasoning is to improve the accuracy-efficiency frontier. However, jointly improving reasoning accuracy and inference efficiency can be challenging, as the two objectives can favor different reasoning behaviors. Independently post-trained models already offer distinct strengths in accuracy and efficiency. We introduce Lightning Weave, a post-training framework that extracts and composes these independently learned capabilities in a single student through on-policy distillation. Each acquired capability is represented by the policy shift from the model before post-training to the resulting specialist. Lightning Weave combines aligned log-ratio shifts at shared student token states and uses Tilted-Target DOPD to convert the cached signals into a stable learning target. Each anchor pair scores the cached trajectories once, enabling subsequent student training without serving multiple live anchor models concurrently. Across diverse student models and benchmarks in mathematics and code, Lightning Weave substantially improves upon the base students and achieves a state-of-the-art accuracy-efficiency frontier. On Qwen3.5-4B, it raises HMMT 2025 accuracy from 59.2% to 64.0% with 10.7% fewer response tokens, and LiveCodeBench v5 accuracy from 41.7% to 54.2% with 9.6% fewer response tokens. Adjusting the relative strengths of the anchor signals yields a strong empirical accuracy-efficiency Pareto frontier. These results establish Lightning Weave as a new practical route to efficient reasoning through capability composition. Code is released at https://github.com/jet-ai-projects/Lightning-Weave.

cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36361

HELICS: Biobank-scale Conditional Synthetic Genome Generation via Latent Flow Matching

Ahmad Abdel-Azim, Xihong Lin

原文摘要

Individual-level genotype data underpin genetic discovery, disease-risk modeling, and therapeutic development, yet many critical settings remain data-limited, including underrepresented ancestries, rare or imbalanced disease cohorts, and restricted-access biobanks. Conditional synthetic genome generation would support privacy-aware benchmarking, simulation, and data augmentation, but requires learning an ultra-high-dimensional discrete distribution over hundreds of thousands to tens of millions of correlated single-nucleotide polymorphisms (SNPs), or sites of DNA variation. We introduce HELICS: High-fidelity gEneration via Latent Interpolant Conditional Synthesis, a phasing-free latent flow matching framework for genome-wide SNP-level generation from biobank-scale genotype data. HELICS tokenizes chromosomes with convolutional autoencoders, reducing >600K SNPs to 4,096 continuous latent tokens, and trains a transformer parameterized conditional flow matching model over the concatenated multi-chromosome latent representation. Trained on over 417K genomes in the UK Biobank, HELICS generates genome-wide common-variant synthetic cohorts without haplotype phasing and achieves stronger fidelity to real genotypes than existing haplotype-based simulators, including high allele-frequency agreement in external evaluation and superior preservation of local and long-range SNP correlation structure. Conditioning on ancestry and genetic liability summaries enables targeted generation across genetic risk profiles; in perturbation analyses, by increasing the conditioned disease trait liability, HELICS induces SNP-level dosage changes in generated genomes that are highly correlated with published association weights. HELICS provides a scalable generative modeling framework for SNP-level genome data and a path toward conditional synthetic cohorts for data-limited genetic studies.

q-bio.GN · cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36360

AK_QMUL at ArGuard Shared Tasks: A Single Fine-Tuned JAIS for Arabic Harmful-Prompt Detection

Ali Khairallah, Arkaitz Zubiaga

原文摘要

We present AK_QMUL's submissions to the ArGuard-2026 shared tasks on Arabic harmful-prompt detection: a binary safe/unsafe task (B1) and a seven-way harm-domain task (B2). Both submissions use one backbone -- JAIS-6.7B fine-tuned with 4-bit QLoRA and a linear classification head over its last-token representation. For B1 we train a four-way head that factors the label by dialect register {safe, unsafe} x {MSA, other}, we augment the training fold with MSA prompts from AraSafe, and decode with a plain argmax to score safe/unsafe. For B2 we fine-tune the same backbone as a seven-way classifier and average the softmax outputs of five random seeds. Both submissions ranked first on their track, with macro-F1 0.984 on B1 and 0.790 on B2. We report results by dialect and by class: on B1 the model is accurate across all six dialects (93.6--99.8%) and its predicted unsafe rate tracks the true rate closely; on B2 six of the seven classes reach F1 between 0.73 and 0.98, with the residual error concentrated in the Bully class, which overlaps semantically with harassment and harm-to-others.

cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36355

WAKMan: Workload Aware KVCache Manager

Suhas Kumar Morisetty, Vaibhavi Lokegaonkar, Sachin Mathew Varghese, Wesley Harris

原文摘要

Prefix caching enables a Large Language Model (LLM) servers to reuse the Key-Value (KV) states and accelerate prefill computation stage for requests that share a prefix with earlier prompts, thus reducing latency and serving cost. With limited memory resource, the KV cache eviction policy becomes a decisive factor in determining the overall benefits of prefix caching. Production traffic may have single-turn, multi-turn, agentic, or a mixture of any combination of these request types. Our findings indicate that agentic steps return within seconds and favor recency, while popular prompts and long conversations return after many other requests and favor frequency. Across 20 mixes, no single policy is always best: vLLM's default strategy LRU provides 12% higher cache hits as compared to the frequency-oriented S3FIFO on agentic-heavy mixes but trails it by up to 22% elsewhere, and even a single workload parameter can flip the winner even for any particular mix. We present WAKMan, a workload-aware KV cache manager that picks its eviction policy from live traffic without knowing request types or reading prompts. It improves hit rate over LRU by 5 percentage points on average, with the lowest mean and worst-case optimality gap of all policies.

cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36353

Detect Before You Disambiguate: Perturbation-Based Input Ambiguity Detection in LLMs

Tiejin Chen, Xiaoou Liu, Michael Clifford, Ahmadreza Moradipari, Yitao Chen, Kyungtae Han, Nejib Ammar, Hua Wei

原文摘要

Large language models (LLMs) often answer ambiguous inputs as if they had a single intended meaning. This is problematic because input ambiguity is an upstream source of uncertainty: when a prompt admits multiple valid interpretations, the key question is whether the model should answer directly at all. Existing input uncertainty methods often rely on LLM-generated statements or clarifications, but such transformations may already disambiguate the input before the ambiguity is measured. We propose a label-preserving perturbation framework for input ambiguity detection in which perturbations remain close to the original input while preserving its label. We instantiate this principle with two detectors: a black-box rule-based perturbation library and a white-box importance-guided method using model-internal signals. Experiments across four datasets in open-domain, conversational, and visual QA show that our methods outperform output-side uncertainty baselines and LLM-paraphrase-based perturbation baselines. Our results highlight the importance of detecting ambiguity before attempting to disambiguate it. Our code is available \href{https://github.com/Xiao0o0o/input-ambiguity-detection}{\textcolor{purple}{here}}.

cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36351

On the Fundamental Limits of Bistatic ISAC Over Vector Channels

Dongsheng Peng, Lei Yu, Ping Chen, Jun Chen

原文摘要

This paper studies the fundamental limits of bistatic integrated sensing and communications (ISAC) systems over vector channels. We first consider a class of parallel channels in which, on each component channel, the channel from the transmitter (Tx) to the communication receiver (Rx) is degraded with respect to the channel from the Tx to the sensing Rx, or vice versa, where the direction of degradation may vary across component channels. For this class of channels, we establish matching achievability and converse results, which yield a single-letter characterization of the capacity-distortion function $C(D)$. For parallel Gaussian channels with additive Gaussian states, the evaluation of the capacity-distortion-power function $C(D,P)$ under the squared error distortion measure reduces to a joint power and distortion allocation problem across the component channels. We then extend the analysis to general vector channels and derive single-letter lower and upper bounds on $C(D)$. In particular, for vector Gaussian channels with additive Gaussian states, we prove the optimality of Gaussian superposition coding without requiring a global degradation order between the communication and sensing channels, by combining the channel enhancement technique with a matrix extremal inequality based on Fisher information. This leads to an exact characterization of $C(D,P)$.

cs.IT

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2602.17700

MIDAS: Mosaic Input-Specific Differentiable Architecture Search

Konstanty Subbotko

原文摘要

Differentiable Neural Architecture Search (NAS) provides efficient, gradient-based methods for automatically designing neural networks, yet its adoption remains limited in practice. We present MIDAS, an approach that modernizes DARTS by replacing static architecture parameters with dynamic, input-specific parameters computed via self-attention. To improve robustness, MIDAS (i) localizes the architecture selection by computing it separately for each spatial patch of the activation map, and (ii) introduces a parameter-free, topology-aware search space that models the two-edge node topology during search. We evaluate MIDAS on the DARTS, NAS-Bench-201, and RDARTS search spaces. In DARTS, it reaches 97.42% top-1 on CIFAR-10 and 83.38% on CIFAR-100. In NAS-Bench-201, it discovers optimal or near-optimal architectures. In RDARTS, it achieves the best reported performance among the compared methods on two of four search spaces on CIFAR-10. We further analyze why MIDAS works, showing that patchwise attention improves discrimination among candidate operations and that the resulting input-specific parameter distributions are class-aware.

cs.LG · cs.AI · cs.NE

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.08746

The Global Empirical NTK: Self-Referential Bias and Dimensionality of Gradient Descent Learning

James Hazelden, Laura Driscoll, Eli Shlizerman, Eric Shea-Brown

原文摘要

In training a neural network with gradient descent (GD), each iteration induces a linear operator that governs the first-order updates to a model's internal state variables. We call this operator the Global Empirical Neural Tangent Kernel (GeNTK). In finite-width networks, GeNTK is typically intractable to form, leading prior work to focus on restrictive settings, such as tracking only outputs or infinite-width limits. Here, we bridge this gap by studying the structure of GeNTK across many models. Formulating the model state as the solution to a global implicit constraint, we derive GeNTK as a product of two operators: K, accounting for immediate parameter-to-state interactions, and P, describing internal state-to-state dependencies. For a broad class of "weight-based models", such as RNNs and transformers, we prove a universal Kronecker-core theorem showing that K admits an exact, forward-pass-computable form given by the Gram matrix of weight-site variables. This core gives a measurable state-dependent restriction on Euclidean GD, termed "self-referential bias": current activity filters which adjoint directions can drive a first-order update. For recurrent models (GRUs and RNNs), we examine the spectrum of GeNTK and formalize notions of low-rank, biased learning in terms of reduced spatial or temporal ranks of the full operator. Practically, this explains why some RNN initializations yield better learning, while some plateau, by biasing the GeNTK and thus restricting learning to a small domain of learnable tasks. To demonstrate computational viability, we validate the main theorems in a self-attention transformer and a pretrained Pythia transformer language model with 410 million parameters. Overall, our results expose tractable local rank and mode-selective biases in the finite-width GD geometry and open the door to principled study of the GeNTK, a bridge between theory and practice.

cs.LG · math.DS · math.OC

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36349

A Local Stabilized Isogeometric Formulation with Discontinuous Pressure for Mesh-Conforming and Immersed Analysis of Nearly Incompressible Elasticity

Ashkan Ali Madadi, Thomas J. R. Hughes

原文摘要

Nearly incompressible materials arise in soft tissues, elastomers, gels, and other soft solids. Yet, numerical modeling remains challenging because displacement-based formulations can exhibit volumetric locking near incompressibility. Mixed displacement-pressure formulations address this difficulty but require compatible approximation spaces satisfying the inf-sup condition. In mixed formulations, discontinuous pressure spaces are attractive since pressure degrees of freedom can be condensed locally. In high-order isogeometric analysis, however, their Galerkin stability requires pressure-grid coarsening that increases with polynomial degree. Here, we develop a variational multiscale (VMS) stabilized mixed isogeometric formulation that retains this locality while substantially relaxing the restriction. The key observation is that VMS stabilization controls nonconstant elementwise pressure modes and is unconditionally stable with respect to its parameter, provided the elementwise constant pressure component satisfies the inf-sup condition. Consequently, for maximally continuous displacement splines, a pressure grid coarsened by a factor of two satisfies the constant-mode inf-sup criterion independently of polynomial degree. Local condensation then yields a displacement-only system, facilitating use in existing displacement solvers. The formulation is further extended to immersed isogeometric analysis with extended B-splines, pressure-cell aggregation, subdivision-based cut-cell quadrature, and weak enforcement of essential boundary conditions. Discrete inf-sup studies and convergence analyses in the Kirsch and Cook's membrane benchmarks demonstrate stability and show that the immersed treatment preserves mixed stability properties. The proposed framework therefore provides a local, high-order, and robust approach for nearly incompressible problems in mesh-conforming and immersed settings.

cs.CE · math.NA

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36348

Thermodynamic characterization of message-passing: Framework and initial study

Kalana G. Abeywardena, Stark C. Draper

原文摘要

We study the thermodynamic efficiency of message-passing decoders using the mismatch cost framework. This framework quantifies the excess entropy generated when a system operates away from the thermodynamically optimal initial distribution. It provides a computable lower bound on irrecoverable energy loss. The decoder is modeled as a main decoding program with local message updates implemented as subroutines. Memory is treated as an external copying subsystem whose thermodynamic cost is excluded. Studying the $[7,4]$ Hamming code transmitted over a binary erasure channel, we show how the channel-induced initialization statistics and the step-wise evolution of the algorithmic state distribution during decoding determine the total mismatch cost. We further show that different thermodynamically optimal state distributions, representing different underlying physical implementations of the decoder, can produce different thermodynamic costs for identical decoding dynamics. Conditioning the input distribution on correctable and non-correctable erasure patterns further reveals distinct thermodynamic signatures. These results demonstrate that thermodynamic cost is jointly governed by decoding dynamics, input statistics, and hardware characteristics. This points to opportunities for thermodynamically-informed decoder design to realize energy-efficient physical implementations.

cs.IT

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36347

Callosal: Rethinking the Kernel Page Cache for Disaggregated Storage and Memory

James McMahon, Amit Samanta, Seung-seob Lee, Ryan Stutsman, Anurag Khandelwal

原文摘要

Classic distributed filesystems like NFS centralize filesystem operations on a server and use custom, filesystem-specific approaches for managing cache consistency across client nodes. This limits scalability, and it leads to weak consistency, ruling out applications that concurrently access and mutate shared files. Our key insight is that these legacy designs are now too storage-centric; with today's large memories, OS page caches do most of the "heavy lifting" in accessing data, not the filesystem and block storage layers. In response to this, we present Callosal, a page cache for a cluster of machines that is centered on shared disaggregated memory. Rather than serializing operations through a filesystem server and using ad hoc caching like NFS, Callosal is built around the existing Linux page cache paths and is accelerated by RDMA and programmable switches. Callosal delivers the strong consistency that programs expect with local filesystems (like in ext4 and tmpfs) while avoiding the additional network access delays and bottlenecks that host-based solutions incur. We compare Callosal against NFS, CephFS, and local filesystems; often Callosal can provide caching performance similar to local filesystems while also scaling across nodes. For mixed read/write workloads, on average Callosal has 4.6x higher throughput than NFS when contention is not extreme, and it offers strong consistency guarantees like CephFS, while outperforming it. For a file-backed key-value store with 1% writes and uniform access, Callosal scales 1.7x across 4 nodes, while CephFS collapses across nodes.

cs.OS

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.36346

Your Wish Is My Commit: Over-Compliant Refactoring in AI Coding Agents

Haocong Jiang, Sadra Sabouri, Sepand Haghighi, Souti Chattopadhyay

原文摘要

Large language models are known to be sycophantic, over-complying with users even when doing so is unproductive. When these models act as coding agents, this can be costly: an agent that keeps editing after no edit improves the code writes churn, technical debt, and review burden into the codebase. Yet the consequences of sycophancy in software engineering remain largely understudied. To understand them, we run a controlled study of iterative refactoring, an open-ended task with no stopping criterion, where any edit an agent makes once the code is already clean reflects compliance rather than improvement. Across algorithmic (n=50) and object-oriented (n=10) suites, we repeatedly ask two widely used agents, Codex and Claude Code, to refactor further across five models. Over 3,000 turns, we define six behavioral signals of over-compliance. We hypothesized that a capable agent would stop once it recognized an optimized codebase; none did. Every model kept editing, rewriting up to 31x the original codebase before first stopping and resuming edits even after declaring the task complete. Over-compliance is a costly, multi-turn behavior that compounds with every turn, and our six signals give developers and tool builders a way to detect and curb it.

cs.SE · cs.HC

查看论文
arXiv ↗PDF ↗