跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+1022新增论文
+813其中 AI 相关
+85已有论文修订
+95新增 AI 导读

arXiv 最近完成扫描:2026-11-17 04:58 · 全站最近入库:2026-11-17 13:07各信源进度 →

统计口径与数据新鲜度

统计 2026-11-16 14:44 — 2026-11-17 14:44(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:2026-11-17 04:58最近发现新增版本:2026-11-17 13:07库内论文最新更新时间:2026-11-16 17:16

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。

24,849 已收录论文20,754 AI 相关7,213 有修订112 HF 社区发现

浏览论文库

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 24,849 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-17 04:57(北京时间)
新论文 · v1arXiv:2611.09014

Did I See It or Think It? Probing Reality Monitoring in Vision-Language Models with Concept Injection

Tina Khezresmaeilzadeh, Seyedarmin Azizi, Mehdi Kamal, Massoud Pedram

原文摘要

Recent work tests whether language models can introspect by injecting a concept direction into their activations and asking whether they notice it and can name it. These studies are text-only. Vision-language models (VLMs) face an additional problem, known in cognitive psychology as reality monitoring: they must keep internally generated content separate from what they actually see. We therefore ask: can VLMs distinguish injected concepts from image content? We introduce a paradigm that varies the image and the injected concept independently. On each trial, the model reports what it sees, whether a thought was injected, and what that thought is. This design scores both whether the model identifies the concept and which source it assigns to it. Across six VLMs, only Gemma-3-27B reliably identifies injected concepts (35% vs. 3% for steering alone), with almost no false alarms. Even then, the source it reports is fragile. Strong or visual-pathway injections are reported as seen, and a visible object nearly eliminates correct thought reports. The same failure appears in ordinary captioning, where injected objects appear in up to 87% of captions of real photographs. Evaluating multimodal self-reports therefore requires checking both their content and their reported source.

cs.CV · cs.AI

新论文 · v1arXiv:2611.09013

KnowEvo: Knowledge Evolution for Protein Optimization

Zijie Xing, Xingyue Liu, Runze Wang, Luoming Hu, Yanming Shen

原文摘要

Large language models (LLMs) have shown promise in scientific discovery, yet systematically accumulating and refining reusable scientific knowledge remains challenging. Existing approaches primarily improve LLM-based problem solving through parameter optimization or feedback-driven refinement of prompts, programs, and solution trajectories, leaving acquired expertise either embedded in model weights or scattered across iterative problem-solving processes. As a result,knowledge is difficult to systematically inspect, evaluate, and refine. We propose KnowEvo, a framework that treats externalized knowledge itself as the optimization target. Our key insight is that scientific knowledge is inherently modular: multiple interacting mechanisms jointly determine the design outcomes. This makes knowledge evolution fundamentally challenging, as improvements cannot be attributed to individual components in isolation. To address this, KnowEvo represents knowledge as executable expertise blocks and organizes their refinements into strategy mutation trees, enabling structured exploration over a combinatorial knowledge space. The framework combines a Thompson-style forward allocation strategy that adaptively distributes evaluation budgets across strategy variants with a backward revision process, in which a code agent operates in an isolated sandbox to translate rollout evidence into localized program updates. This design enables persistent accumulation, evaluation, and refinement of mechanistic expertise across iterative experimentation. We evaluate KnowEvo on protein optimization tasks including stability and solubility. Results show that knowledge evolution consistently improves over initial strategy blocks and outperforms monolithic program evolution and prior LLM-guided evolution baselines across metrics such as pass@k, MRR, MAP,and NDCG@5. Code is available at https://github.com/recynie/KnowEvo.

cs.LG

版本更新 · v3arXiv:2610.38026

Making Sense of Audience Change: Causal Attribution and Boundary Management on Xiaohongshu

Erin Zhou, Kyrie Zhou

原文摘要

Recommendation-driven platforms can reshape not only what users see, but also who becomes visible and how community boundaries are negotiated. We examine Xiaohongshu discussions surrounding a perceived World Cup-related audience influx to understand how users attributed community change and how those explanations related to boundary-management strategies. We analyzed 262 comments from eight high-engagement threads using researcher-led thematic analysis, structured coding, and permutation-based association analysis. Users described personalization as both social separation and a boundary that could fail, interpreted platform growth through audience acquisition and commercialization, and contested whether gender should define community membership. Among 70 comments with explicit attributions and strategies, attribution and strategy were strongly associated (Cramer's V = .583): algorithmic attributions concentrated in technical boundary management, while social-actor attributions more often aligned with social enforcement or normative socialization. These findings connect causal sensemaking with boundary management and show how audience restructuring can create governance consequences on recommendation-driven platforms.

cs.HC · cs.AI · cs.CY

新论文 · v1arXiv:2611.09012

Towards Robust Reinforcement Learning Market Making: Lessons from Loss Function Experimentation

Siddharth Nand, Bora Guney

原文摘要

Market makers play a vital role in ensuring efficient trade within financial markets. Recent literature has delved into training market-making systems using reinforcement learning. This paper delves into the impact of various loss function formulations on risk management behavior within reinforcement-based market making. Through empirical evaluation, we contrast profit-maximizing strategies with risk-averse approaches, finding that penalties for inventory and cash management foster more resilient and adaptable market-making strategies. Our results underscore the significance of striking a balance between profit maximization and risk management in algorithmic trading, providing actionable insights for refining strategies in practical settings.

cs.LG · q-fin.CP · q-fin.TR

版本更新 · v2arXiv:2506.09661

A Cytology Dataset for Early Detection of Oral Squamous Cell Carcinoma

Garima Jain, Sanghamitra Pati, Mona Duggal, Amit Sethi, Abhijeet Patil, Gururaj Malekar, Nilesh Kowe, Jitender Kumar 等 37 位作者

原文摘要

Oral squamous cell carcinoma OSCC is a major global health burden, particularly in several regions across Asia, Africa, and South America, where it accounts for a significant proportion of cancer cases. Early detection dramatically improves outcomes, with stage I cancers achieving up to 90 percent survival. However, traditional diagnosis based on histopathology has limited accessibility in low-resource settings because it is invasive, resource-intensive, and reliant on expert pathologists. On the other hand, oral cytology of brush biopsy offers a minimally invasive and lower cost alternative, provided that the remaining challenges, inter observer variability and unavailability of expert pathologists can be addressed using artificial intelligence. Development and validation of robust AI solutions requires access to large, labeled, and multi-source datasets to train high capacity models that generalize across domain shifts. We introduce the first large and multicenter oral cytology dataset, comprising annotated slides stained with Papanicolaou(PAP) and May-Grunwald-Giemsa(MGG) protocols, collected from ten tertiary medical centers in India. The dataset is labeled and annotated by expert pathologists for cellular anomaly classification and detection, is designed to advance AI driven diagnostic methods. By filling the gap in publicly available oral cytology datasets, this resource aims to enhance automated detection, reduce diagnostic errors, and improve early OSCC diagnosis in resource-constrained settings, ultimately contributing to reduced mortality and better patient outcomes worldwide.

eess.IV · cs.CV · q-bio.TO

新论文 · v1arXiv:2611.09008

ISER-MoE: Dynamic In-Switch Replication of Hot Experts for Load Balancing in Expert-Parallel

Jintao Peng, Ethan Zhang, Kefeng Duan, Bangyu Sheng, Min Li, Itamar Rosenfeld Rauch, Jack Yang, June Yang

原文摘要

Expert-parallel mixture-of-experts (MoE) layers run at the pace of the most loaded GPU, and a few hot experts carry most of the excess load. Replicating them at every MoE layer balances the load only if their weights arrive within the layer, and broadcasting from the owner is the bottleneck: point-to-point copies overload it, and baseline broadcasts pay latency or bandwidth overheads. We present ISER-MoE (In-Switch Expert Replication for MoE), built in TensorRT-LLM, which multicasts hot experts through the NVLink switch to fixed GPU groups, so each byte leaves its owner once. A GPU scheduler assigns hot experts to groups per layer, and a multi-port broadcast spreads the injection over the group's ports, both overlapped with the shared expert. In trace replay, the scheduler removed 98-99% of the excess load; in prefill serving of DeepSeek-V4-Pro, throughput rose 17.4% on a math benchmark and 7.2% on agentic prompts.

cs.DC

新论文 · v1arXiv:2611.09006

Rethinking Latent Reasoning for Generative Recommendation: Beyond Teacher-Generated CoT and Recurrent Hidden-State Feedback

Yiwen Chen, Fuwei Zhang, Zehao Chen, Deqing Wang, Hehan Li, Peizhi Xu, Hanmeng Liu, Shuanglong Li 等 11 位作者

原文摘要

Latent reasoning in generative recommendation faces two practical barriers: constructing teacher-generated chain-of-thought (CoT) supervision is costly, and recurrent hidden-state feedback requires a specialized execution loop. We revisit both the supervision source and the computation interface. A step-level diagnosis measures how recommendation CoT supports next-item prediction and trace coherence. In the analyzed traces, history analysis is the only step with positive mean contributions on both dimensions, while later preference and intent steps have negative mean answer contributions. Guided by this finding, we propose PHiT (Prefilled, History-Informed Thought Tokens), which learns intermediate states from semantic targets extracted directly from observed item histories. PHiT aggregates cached sparse item encodings to construct these targets and supervises the contextual states of distinct vocabulary thought tokens. All thought tokens are supplied together as ordinary input IDs during a single causal prefill, removing the need for recurrent final-state feedback. With supervised fine-tuning alone, PHiT outperforms the compared baselines across four recommendation benchmarks. Controlled ablations support the value of input-matched semantic supervision, while cost measurements show fast target construction with training time comparable to the recommendation-only thought-token baseline.

cs.IR

版本更新 · v2arXiv:2610.26165

TriTD: Tri-Partite Trajectory-Distribution Distillation for Real-Time Autoregressive Video Generation

Jia Li, Xurui Peng, Haowei Zhu, Tianyu Zhao, Jiexi Wang, Youwei Zheng, Yuxi Ren, Xinglong Wu 等 10 位作者

原文摘要

Autoregressive (AR) video diffusion has emerged as a leading paradigm for efficient video generation, yet its practical deployment remains bottlenecked by the trade-off between generation quality and inference efficiency. We attribute this bottleneck to three error sources in AR rollouts: the base model error, the train-inference distribution mismatch, and the upstream error propagation. Building on this decomposition, we derive closed-form upper bounds on the per-step and cumulative errors, providing a principled basis for jointly optimizing generation quality and inference efficiency. Guided by this analysis, we propose TriTD, a Tri-Partite Trajectory-Distribution co-optimized distillation framework with three complementary modules. First, for ODE initialization, we adopt a Next-Node Trajectory Distillation (NTD) objective, which shortens the supervision horizon and admits a lower regression-error floor. Second, we introduce Chunk-Graded Noising (CGN), a monotonically non-decreasing per-chunk noising schedule that helps narrow the expected train-inference noise mismatch and removes redundant KV-cache updates, yielding substantial speedups at no quality cost. Third, we propose Orthogonal-Parallel DMD (OP-DMD), a distribution-matching loss that retains the original DMD supervision while introducing an additional regression constraint in the orthogonal subspace to suppress off-axis drift, improving both training stability and final generation quality. Extensive experiments show that TriTD consistently surpasses existing baselines in visual quality while reaching a real-time throughput of 26.8 FPS on a single H100 GPU, delivering a balanced quality-efficiency solution for interactive streaming video generation.

cs.CV

版本更新 · v2arXiv:2606.09355

MosaicIMU: Composing Carrier Experts for Generalizable Neural Inertial Odometry

Junye Zou, Huiyi Yan, Xinning Xu, Xiaolei Li, Pengkun Zhou, Jinhui Zhang, Ziyang Meng

原文摘要

Robust inertial odometry is essential for various carriers when external sensing is unreliable. Learning-based methods reduce integration drift by capturing local motion priors, but these methods often remain tied to a particular carrier, limiting generalization across heterogeneous platforms. We present MosaicIMU, a carrier-conditioned Mixture-of-Experts (MoE) pretraining-and-adaptation framework for generalizable neural inertial odometry. MosaicIMU uses a prototype-based router to compose carrier-specific expert features, decodes local velocity and uncertainty constraints, and integrates them with a history-aware EKF. For unseen domain adaptation, it freezes the pretrained base model and learns a new lightweight expert residual branch. For edge-deployment, it further reuses the router to select informative online samples for efficient incremental updates. Experiments show that MosaicIMU consistently outperforms learning-based baselines, reducing average ATE and RTE-10s by 40% and 34%, respectively. These results highlight that MosaicIMU provides a scalable pretraining-to-deployment paradigm for generalizable and adaptive neural inertial odometry.

cs.RO

新论文 · v1arXiv:2611.09005

FASTFUNNELS: Agile Multi Robot Navigation in Narrow Spaces via Adaptive Dynamic Funnels

Vignesh Rajagopal, Rohan Chandra

原文摘要

We focus on the problem of high-speed multi-robot navigation in cluttered spaces with tight spatial constraints. A key challenge is balancing safety with agility, as existing methods often trade one for the other (slowing down near a bottleneck to avoid collisions versus failing to swerve in time at high speeds), especially for car-like systems. Our approach belongs to the family of formation control methods, in which Ackermann-steering agents coordinate while navigating tight spaces. However, current formation control techniques rely on rigid, pre-defined shapes and transitions that depend on initial conditions, the environment, and vehicle dynamics, and are impractical to maintain at high speeds. For example, agents passing through a small gap must first form a square, transform into a line using analytical equations, and then return to a square. In contrast, our approach is inspired by a droplet of water that continuously adapts its shape to the space it flows through without slowing down. We propose a hierarchical algorithm in which, at the high level, a reinforcement learning policy produces a dynamic, blob-like elliptical patch that morphs its shape in real time, squeezing, stretching, and orienting itself around obstacles and through tight spaces. At the low level, the robots are controlled via distributed model predictive control (DMPC) to stay inside the patch, which acts as a soft forcing function, while avoiding collisions with each other. We evaluate our approach, FastFunnels, on simulated F1TENTH racing cars across 14 scenarios with N=1 to N=6 robots, averaged over 20 trials each. FastFunnels achieves an 89% completion rate at 7.73 m/s, while all baselines either fail to complete the track or succeed only at very low speeds (0.07-1.79 m/s, about 40% completion rate).

cs.RO

新论文 · v1arXiv:2611.09004

Sharing the Road: LLM-Based Bargaining in Mixed-Autonomy Traffic

Lingxiao Zhou, Di Chen, Jia Li

原文摘要

Mixed-autonomy traffic involves heterogeneous driving populations competing for shared road space. Previous studies have shown the existence and emergence of collective rationality using game-theoretic and reinforcement learning approaches, which rely on explicitly specified behavioral assumptions or reward functions for driving agents. However, how bargaining behaviors between driving agents evolve and lead to collective outcomes based on richer agent characteristics and language-based reasoning remains less understood. To fill this gap, we develop a bargaining framework based on large language models (LLMs) in which population-level autonomous vehicle (AV) and human-driven vehicle (HV) agents negotiate road-space allocations under heterogeneous objectives and private information. Through controlled experiments varying traffic composition, agent objectives, and traffic-domain knowledge, we examine bargaining behavior, coordination, and resource utilization. Results show that negotiated allocations respond to relative traffic density without strictly following proportional sharing, while density-balanced states exhibit greater coordination difficulty. Fairness-oriented objectives substantially improve agreement rates but do not necessarily yield equally high road-space utilization. Furthermore, domain knowledge affects bargaining outcomes differently across objective configurations, shaping both coordination and the role-dependent effects of fairness. These findings demonstrate the potential of LLM-based bargaining for modeling and designing interactions among driving agents in mixed-autonomy traffic.

cs.MA

版本更新 · v2arXiv:2509.23178

Limit Analysis for Symbolic Multi-step Reasoning Tasks with Information Propagation Rules Based on Transformers

Tian Qin, Yuhan Chen, Zhiwei Wang, Zhi-Qin John Xu

原文摘要

Transformers are able to perform reasoning tasks, however the intrinsic mechanism remains widely open. In this paper we propose a set of information propagation rules based on Transformers and utilize symbolic reasoning tasks to theoretically analyze the limit reasoning steps. We show that the limit number of reasoning steps is between $2^{L-1}-1$ and $\frac{3^{L-1}-1}{2}$ for a model with $L$ attention layers in a single-pass. In addition, we find that, for problems with a fixed number of reasoning steps, Transformers exhibit a preference for generating sequences that satisfy certain ordering relations. Ultimately, we construct a Transformer and prove that the limiting reasoning depth is achievable.

cs.AI

版本更新 · v2arXiv:2611.07009

TraCE: Trajectory-Aware Concepts for Explainable Video Understanding

Wooil Lee, Junpyo Hong, Jongseo Lee, Tuomas Oikarinen, Tsui-Wei Weng, Jinwoo Choi

原文摘要

Video understanding models often rely on motion, yet their explanations can obscure which motion sources support a prediction. We identify motion-source under-specification as a limitation of existing video XAI: region-centric explanations can be source-ambiguous, while motion vocabularies restricted to single-person pose sequences are source-incomplete. We propose TraCE (Trajectory-Aware Concepts for Explainable Video Understanding), a concept-based framework that discovers source-flexible motion concepts by clustering optical-flow trajectories from training videos. These concepts represent dynamic evidence from people, objects, and interactions, while object and scene concepts account for static context. The resulting typed trajectory/object/scene interface supports class prediction for action recognition and question-conditioned concept matching for video question answering (VQA). We evaluate TraCE across five random seeds on six benchmarks: NTU RGB+D Mutual, Chiral SSv2, UCF-101, KTH, TOMATO, and STAR. Relative to non-interpretable baselines, TraCE maintains competitive task performance, with an average accuracy drop of 1.0 points for action recognition and a nominal gain of 0.4 points for VQA. Motion perturbation and concept-removal tests provide complementary evidence of directional semantic alignment and decision faithfulness, while sample-level interventions illustrate targeted model debugging. Our project page is available at https://leewooil.github.io/TraCE_project_page/.

cs.CV

新论文 · v1arXiv:2611.09003

FoundaRF: A Physics-Informed Foundation Model Framework for Radio-Frequency Passive Components

Houbo He, Chris Han, Hyunsu Chae, Lei Xia, Qiang Zhou, Lawrence Liu, Sensen Li, David Z. Pan 等 9 位作者

原文摘要

Designing RF/mmWave passive components is a time-consuming process that relies heavily on expert knowledge and iterative EM simulations. Although recent ML surrogate models have shown great promise in accelerating passive design, they remain task-specific and lack generalization. This paper presents FoundaRF, a foundation model framework that unifies passive layout modeling across diverse layout structures and technologies. It combines a self-supervised encoder with adaptive layout processing and physics-aware decoders to enable few-shot transfer. Experiments across four semiconductor technologies show strong generalization to unseen structures, outperforming baselines with up to a 3.3x reduction in prediction error, reducing labeled-data requirements by up to fourfold, and substantially shortening design turnaround. FoundaRF establishes a scalable and data-efficient solution for RF/mmWave passive design automation.

eess.SP · cs.AI

新论文 · v1arXiv:2611.08999

Capital Efficiency, Isolation, Risk, and Ruin in Generative AI Inference Serving Infrastructure

Renée St. Amant, Anjaly Parayil, Ankur Mallick, Victor Rühle

原文摘要

Generative AI serving is an extremely capital-intensive endeavor that demands rapid, step-change improvements in capital efficiency to prevent industry wide increases in the cost of capital from limiting growth plans. Keeping pace with the economic necessities will increasingly require infrastructure providers to confront operationally-inconvenient realities of production serving, including the extreme heterogeneity of generative AI request costs. We have explored workload isolation as a foundational, fleet-level lever for increasing capital efficiency. Isolation is the practice of intelligently grouping "like" requests before execution. Rather than exposing every replica to the full heterogeneous workload distribution, isolation deliberately reshapes the workload presented to the hardware by assigning requests with different cost characteristics to different portions of the fleet. Workload isolation on its own does not make expensive requests cheaper; it prevents a costly portion of the workload from unnecessarily determining the operating point and capital efficiency of the entire fleet. It additionally lays the foundational groundwork that can serve as a launching point for compounding gains via narrow-band optimizations (tailoring) and custom silicon designs while also informing up on pricing policy. This paper uses representative production data and experience to convey the motivation behind isolation as a fleet-level lever for increased capital efficiency.

cs.DC · cs.AI

版本更新 · v5arXiv:2606.06840

Characterize Then Distill: Mechanistic Reasoning in Large Output Spaces

Debjyoti Saha Roy

原文摘要

Reasoning-trained language models can select, zero-shot, a small set of relevant labels from thousands to hundreds of thousands of candidates. We ask how they do it mechanistically, and whether the mechanism can be distilled. We treat each decision as a token-level event scored by the model's own margin: the token that picks a coarse region of the label space, the tokens that pick a label within it, and the token where the output departs from a near-miss named earlier. Attribution, exact mean-ablation, knock-in and a null calibration give individual attention heads causal standing. On clinical coding of MIMIC-IV discharge summaries, with all 5,651 candidate codes in context, a small, global, phase-structured set of heads is necessary and sufficient on essentially every summary; distinct head families read the candidate region and the earlier near-miss, and blocking what they read removes their effect; for mentions elicitable in the reasoning, a disjoint mid-layer set makes the region elicitable several tokens before the code. The structure recurs in a second reasoner. We introduce MISTILL, which, unlike chain-of-thought distillation, also supervises the student's pooled attention at these decision events. Read on heads found after training, it nearly doubles the causal recovery of the contrastive decision in a cross-family student and adds a small, seed-stable gain in one that already carries most of it, with no detected task difference at matched precision; under matched selection it exceeds a frozen random-head pool on both recovery and task score. On two corpora with human-annotated evidence (one read zero-shot, one in domain), a MISTILL student's decision heads attend to each code's evidence more, and more specifically, than a chain-of-thought-distilled student's. A pre-registered evidence-flip test detected no improvement in evidence-dependent decisions through the tested routes.

cs.CL · cs.AI · cs.LG

新论文 · v1arXiv:2611.08998

Seeing vs. Counting: Complexity Twins Show Where Multimodal LLMs Fail on Diagrams

Yin Liu

原文摘要

When a multimodal LLM miscounts in a diagram, does it fail at seeing or at counting? Existing visual graph and puzzle benchmarks vary visual density and algorithmic hardness together, confounding the two. We pair counting questions about the same drawing that sit in different counting-complexity classes, such as its spanning trees (a determinant) and its simple paths (#P-complete), and use a transcription probe and a text-only control to separate reading from computing. We test Gemini, Gemma and Qwen3-VL models; for the Gemini models, the answer depends on how the model computes. In chain of thought, most of them read small drawings correctly and then fail at counting, at a rate set far more by the size of the answer than by its complexity class (for the primary model, the odds of an exact answer fall 2.38-fold per decade of the answer). Given code and an exact transcription, the primary model counts correctly even on 10x10 grids, and reading becomes the bottleneck: a single view misses horizontal edges, and a second reading of a rotated copy with crop checks reads 86/96 large grids exactly, against 57/96 when the upright reading is simply repeated.

cs.CV · cs.AI · cs.CL

新论文 · v1arXiv:2611.08997

Calibrating your confidence requires thinking of all the ways you could be wrong

Marco Miani

原文摘要

A primary limitation of modern deep learning is the miscalibration of model confidence; traditional architectures fundamentally lack a mechanism to quantify their own uncertainty. Larger models and larger datasets mitigate this issue, but they do not resolve the underlying structural deficiency. Bayesian methods offer a principled uncertainty-aware learning framework through stochastic parameters. Yet, despite the established benefits, the inherent computational overhead often compromises either scalability or quality. This trade-off currently prevents the widespread adoption of Bayesian methods. This Thesis bridges the gap between high-quality and low-complexity by introducing a novel distribution on parameter space: a degenerate Gaussian whose covariance is a projection matrix aligned with the null space of the Fisher information matrix. We demonstrate the scalability of this Bayesian posterior through two numerically efficient algorithms: one for posterior sampling and another for the direct computation of predictive confidence. Empirically, we show State-Of-The-Art confidence calibration and out-of-distribution detection for large models. We provide theoretical grounding for this approach. We study the degeneracies inherent in Amari's information geometry and highlight its most relevant component: the split of parameter's perturbations into two orthogonal subspaces. From this perspective, our posterior naturally emerges as the simplest choice that captures this split, assigning unitary variance to one subspace and zero variance to the other. Moreover, through the tools of differential geometry, we utilize this split to identify and rectify a critical misalignment issue of the non-linearized Laplace Approximation. Finally, with a focus on PAC-Bayes generalization bounds, we pave the way for a deeper understanding of the learning capabilities of overparametrized models.

cs.LG · stat.ML

新论文 · v1arXiv:2611.08996

UniMore: Fine-Grained Supervision for Unified Understanding and Generation

Haoyue Bai, Yinyi Luo, Yicheng He, Qingsong Wen, Sharon Li, Jindong Wang

原文摘要

Unified multimodal models combine image understanding and generation within one system, creating an opportunity to improve both capabilities through paired image--text supervision. We examine the outputs of a pretrained unified model in both directions and observe a loss of specificity: generated images can contain the requested objects with incorrect attributes, while captions identify a scene without expressing its distinguishing details. These failures call for supervision that specifies both the visual content to realize and the descriptive differences to distinguish. We propose UniMore, a post-training method that derives these complementary learning tasks from the fine-grained content of image--text pairs. The generation route predicts grounded crops from their corresponding phrases, matching the scope of the condition and visual target while using the native visual objective. The understanding route combines positive-caption fitting with ranking against captions that omit or alter details, conditioned on the full image. The two routes are optimized independently, and their learned updates are combined into a single unified model without changing its inference procedure. Experiments show gains across multiple understanding and generation benchmarks.

cs.MM · cs.CV

新论文 · v1arXiv:2611.08995

Same Facts, Different Remedy: Linguistic Profiling and Arbitrariness in LLM and System One Decisions

Colin Priest

原文摘要

Large language models are increasingly used to resolve consumer complaints, and the way complainants write varies with age, education and socioeconomic status (SES). We audit three remedy-decision systems using sociolinguistic rewrites of US consumer financial complaints that are instructed to preserve every fact: two generative LLMs that return a single label (GPT-4o-mini and GPT-5-mini) and a probabilistic "System One" model that returns a distribution over remedies (Jev). Using pre-registered hypotheses tested once on held-out complaints, we separate two phenomena that single-label audits conflate. At the group scale, older, college-educated and higher-SES styles receive slightly more generous remedies; these effects replicate but are small, with every monetary-remedy gap within 0.5 percentage points. At the individual scale, two style versions of the same complaint receive different monetary decisions 2.6-6.3% of the time. In a variance-component analysis of the probabilistic model, persona components shared across the two rewrite models account for about 1.6% of within-complaint variation. The persona gaps are associated mainly with delivery (vocabulary, grammar, politeness) rather than content. Positive controls show that all three systems respond 49-387 times more strongly to remedy-relevant facts than to style. Rewriting every complaint into one standard style reduced GPT-4o-mini's education and SES gaps by 94% and 87% while retaining fact sensitivity, but reversed two of the probabilistic model's gaps and did not make individual decisions more stable. Independent LLM auditors flagged material fact changes in 3.3% and 11.7% of rewrites, and instability persisted among unflagged rewrites. Fairness audits of LLM decision systems should report group disparities and individual instability as separate outcomes.

cs.CY · cs.AI · cs.CL

版本更新 · v2arXiv:2607.19395

From Trajectories to Prefixes: Reusing Teacher Trajectories via Replayed Prefixes and Online Continuation

Yihan Wang, Zhong Guan, Haoran Sun, Jiale Huang, Likang Wu, Hongke Zhao

原文摘要

Small language models are attractive backbones for interactive agents, but direct distillation from strong teacher trajectories often turns rich multi-turn behavior into one-shot imitation targets. This is inefficient in long-horizon environments, where early decisions shape later states and rewards. We propose Prefix-GRPO, a reinforcement learning framework that decomposes teacher trajectories into replay-aligned prefix queries and online continuations. Each prefix is replayed in the environment to recover a valid intermediate state, after which the student continues online interaction and receives task reward. Unlike response-only GRPO, Prefix-GRPO also applies clipped policy updates to historical assistant tokens inside the replayed prefix, using a policy-distilled SFT checkpoint to estimate their old log-probabilities. This unifies prefix learning and continuation learning within the same policy-optimization form. Experiments on TextCraft, BabyAI, and ALFWorld show that Prefix-GRPO improves small-model agents over distillation and standard RL baselines, while ablations show that replay alone is insufficient without explicit prefix-token optimization. The implementation and reproduction scripts are available at https://github.com/HappynessI/Prefix_GRPO.

cs.LG · cs.AI

版本更新 · v2arXiv:2606.04413

(Mis)generalization of Helpful-Only Fine-tuning

Mohammad Omar Khursheed, Baram Sosis, Fabien Roger

原文摘要

Helpful-only models, that is, models that are trained to always follow user intent, are valuable for dangerous capability evaluations and other areas of AI R&D where refusals would be an obstacle. Little is known about the generalization properties of helpful-only training: helpful-only models refuse less than their harmless counterparts, but previous work has not studied other dimensions of their alignment. We study the shortcomings of existing helpful-only models. We find that some show emergent misalignment, others have residual refusal behaviors, and most show poor steerability, sycophancy, and incoherent character. We show that simple anti-refusal training can cause many of these issues. None of these problems are necessary consequences of helpful-only training, though: we show that synthetic document fine-tuning and adding character-related questions to SFT and RL can mitigate them.

cs.LG

新论文 · v1arXiv:2611.08994

The Drawing You Watch Is the Portrait You Keep: A Replayable Machine Portrait Studio

Vinod Rajpurohit

原文摘要

Ask an image model for your portrait and a finished picture arrives: no first line, no second thoughts, no record of how it got there, and a different picture if you ask again. I built a portrait studio that works the other way. A sitter chooses one of eight fictional artists, each written with a city, a medium and a tempo of its own, holds still for 8 to 12 seconds, and watches a drawing of themselves appear over 44 seconds. One still is sent after consent, a hosted image model returns a private study the sitter never sees, and a compiler turns that study into one drawing document. The live drawing, the saved sheet, the vector export and the film all read that document, and a replay makes no model call, so the drawing you watch is the portrait you keep. I describe the compiler in enough detail to rebuild it and say plainly what it is: an authored reconstruction at a compressed speed, not a recorded hand. On 16 generated test studies a sheet holds 8,784 to 16,378 timed contacts, about three quarters of them fixed-grid fills and nearly all shorter than one video frame, and recompiling one study gave byte-identical output four times on one machine. The paper also traces a bug that matters more than any of that. The earlier compiler estimated the paper colour tile by tile, read dark skin as paper, and in all eight hands kept less of a deep-complexioned face than a fair one (13 to 48 percent against 41 to 56). On 80 generated studies the rebuilt engine keeps the complexion difference the study already has (raw retention 89 to 120 percent), so the fairness question has moved upstream, to the study and the hosted model. The two test sitters are generated and differ in more than complexion, and no real-face or perceptual study exists yet. The studio is live at https://everyfaceproject.com.

cs.HC · cs.CY · cs.GR

新论文 · v1arXiv:2611.08993

When2What: A Timing-Conditioned Framework for Long-Horizon Forecasting of Continuous Cardiovascular Waveforms

Chaoqi Zhang, Shujie Yan, Zhao Zhang, Yanhua Sun, Yu Wang, Li Li, Lingzhong Meng, Xiaoyi Chen 等 9 位作者

原文摘要

Long-horizon forecasting of quasi-periodic cardiovascular waveforms, such as the electrocardiogram (ECG), arterial blood pressure (ABP), and photoplethysmogram (PPG), requires predicting both when future cardiac events will occur (beat timing) and what the local structural pattern looks like around each event (beat morphology). However, standard sequence forecasters trained with pointwise mean squared error (MSE) average across uncertain future heartbeat phases, flattening sharp R peaks and pulse upstrokes into featureless baselines. Our analysis of intraoperative ECG reveals a clear temporal asymmetry: beat-time predictability from history decays within seconds due to stochastic heart-rate variability, whereas local beat morphology aligned at true beat times remains stable over several minutes. Guided by this insight, we introduce When2What, a timing-conditioned framework comprising three stages: Predict, Commit, and Generate. Given a 120-s history, When2What predicts a dense 60-s timing map, selects and sharpens discrete beat events into unit-height Gaussian impulses, and conditions a neural waveform generator on both these committed events and the observed history. Across intraoperative ECG, When2What substantially outperforms existing time-series and physiological forecasters, achieving a near-term waveform event skill of 0.644 and reproducing 95.62% of its predicted heartbeats in the generated waveform where direct long-horizon forecasters collapse into flat baselines. Furthermore, When2What transfers zero-shot to external hospital and atrial fibrillation cohorts, extends effectively to ABP and PPG pulse waveforms, and is supported by ablations and synthetic analyses that establish explicit timing conditioning as an effective paradigm for quasi-periodic long-horizon waveform forecasting.

cs.LG

版本更新 · v3arXiv:2610.01834

Code Owns the Simulation, Jev Owns the Evaluation

Yaodong Yang, Hongyao Tang, Yi Ma, Xingyu Fan, Weixun Wang, Jinpeng Li, Tianpei Yang

原文摘要

Judgment models such as \jev{} return, in a single call and without reasoning text, a probability for each described option. This makes them attractive as an agent's action-selection layer, but it is unclear which decisions they can be trusted with. We test \jev{} on reflection tests, one-shot matrix games, the text game ALFWorld and robot control, and find a sharp boundary. \jev{} succeeds when the right option can be judged from what the input describes, which we call \emph{evaluation}. Specifically, it solves 99\% of the counterintuitive Cognitive Reflection Test questions. However, it fails when the right option depends on \emph{simulation} (i.e., predicting something not in the input), such as the opponent's action or the subgoal that must come first. In games, \jev{} plays suboptimally as if its rational opponent acted at random, because the opponent's action is not given. In ALFWorld, \jev{} favors commands that mention an object or place named in the task description. For example, given the task ``put a clean knife in the drawer'', \jev{} carries an unwashed knife straight to the drawer instead of first washing it at the sink. Surprisingly, many of these failures are not due to a lack of knowledge. Asked separately what the opponent will do, \jev{} usually answers correctly, and it responds well given the opponent's action. It fails when one call must both perform the simulation and evaluate based on it. This suggests letting code make the prediction or simulation. When code supplies it, such as a lookahead in ALFWorld and physics simulation in robot control, \jev{} becomes an expert controller through its general evaluation ability.

cs.AI · cs.LG

新论文 · v1arXiv:2611.08992

Enforcing Symmetries on Pre-trained ML Models for Physics

Wenjie Zhou, Valentino F. Foit, David W. Hogg, Soledad Villar

原文摘要

Post-hoc group averaging (or test-time augmentation) is a simple way to impose symmetry on a trained predictor without changing its architecture or retraining it. For finite and compact groups, averaging over a normalized Haar measure on the group gives the Reynolds operator: a canonical projection onto invariant or equivariant predictors with known non-increasing risk guarantees under certain assumptions. This paper provides numerical experiments on physics emulation models that demonstrate the effectiveness of the approach in this setting. We then extend this framework to non-compact group symmetries. For non-compact groups, the Haar measure cannot be normalized, so group averaging cannot be defined in the same way. We replace it with three alternatives: the Weyl unitary trick (which requires complexifying the space and extending the predictor), a data-dependent orthogonal projection (which involves a non-trivial optimization on the space of invariant or equivariant functions), and a frame averaging approach (which involves mapping each point to a set of orbit representatives). We illustrate the constructions theoretically and empirically on the unit-rescaling group and the Lorentz group. Our results show that in the non-compact case, the Weyl trick and frame averaging do not provide a consistent improvement over the baseline, except where the test data is out-of-distribution (an orders-of-magnitude improvement in this case) and, for the Weyl trick, in low-data regimes. The orthogonal projection can provide improvements in-distribution and out-of-distribution. We also observe that models trained directly on invariant features matched or beat the projection of an unconstrained one. We conclude that post-hoc symmetrization extends to non-compact groups, but it does not consistently outperform training a symmetric model.

cs.LG · physics.comp-ph · stat.ML

新论文 · v1arXiv:2611.08991

Rethinking Optimization Methods for Hierarchical Feature Selection

Xinyuan Song, Zekun Cai, Liang Zhao

原文摘要

Rare-event forecasting often depends on interpretable conjunctions: a single keyword, location, or temporal cue may be weak, while their joint occurrence can identify a meaningful precursor. Hierarchical-Task Numerical conjunctive feature Learning (HTNL) selects Numerical Conjunctive Features (NCFs) through a structured sparsity objective. However, the resulting optimization problem is challenging, making optimization design critical to its practical performance. We give a unified diagnosis of four HTNL optimization families: original majorization-minimization (MM), variational quadratic form (VQF), alternating direction method of multipliers (ADMM), and weighted group lasso (WGL). Our analysis distinguishes three key factors that jointly determine HTNL optimization behavior. First, VQF and MM are algebraically equivalent under matched variational updates and optimize a squared-penalty formulation. Second, ADMM optimizes the unsquared objective through consensus splitting; we replace its truncated FISTA proximal solver with a scalar bisection update (ADMM-Bisect) for more accurate treatment of near-threshold blocks. Third, WGL targets the same unsquared objective, while all methods face rapidly decaying active-set admission scores. Across two synthetic datasets and 11 civil-unrest and flu-surveillance tasks, VQF achieves the highest mean AUC with the most compact non-collapsing supports. ADMM-Bisect reduces measured runtime, while calibrated WGL avoids the empty-support failures of its original scale. Synthetic results further show how proximal accuracy, pruning, and search budget affect candidate recovery, especially for longer conjunctions. Code will be made publicly available upon publication.

cs.AI

新论文 · v1arXiv:2611.08990

Action-Centered Causal Bandits for Budget-Constrained Promotion Targeting

Yifan Lin, Kaixu Yang, Di Mo

原文摘要

Web platforms select among costly interventions for each user to maximize incremental outcomes under a global budget. We formulate this problem as a budget-constrained contextual bandit with heterogeneous treatment effects. Although an action-independent outcome baseline cancels from treatment comparisons, modeling a nonlinear baseline can increase statistical complexity. We propose DR-CATE-CBwK, an action-centered causal bandit that directly learns treatment contrasts. Its online update remains valid under baseline-model misspecification without inverse-propensity weighting. It combines an offline doubly robust learner to warm-start treatment-effect estimates, an online action-centered estimator for sequential contrast learning, and a primal-dual controller with an affordable-action rule to enforce the effective-cost budget pathwise. For linear conditional average treatment effects (CATEs) in a stationary i.i.d. setting, we prove a high-probability (\widetilde{O}(Kd\sqrt{T})) regret bound, where K is the number of priced actions, d the context dimension, and T the horizon. We benchmark against the fluid linear program (LP) under the coverage-guaranteed action rule. We further show that the action-centered estimating moment remains unbiased under any predictable baseline estimate, whereas a specified separate-arm reward regression can exhibit asymptotic contrast bias under action-dependent logging and shared-baseline misspecification. On a semi-synthetic job-marketplace benchmark with real logged features and simulated outcomes, our practical Thompson-restricted variant achieves the lowest regret among evaluated budget-feasible methods in the operating regime. It reduces regret by approximately 20% versus the strongest reward-based baseline, outperforming raw-reward contextual bandits, BwK baselines, and a static offline policy, while never exceeding the effective-cost budget.

cs.LG

版本更新 · v2arXiv:2606.30317

MCP Server Architecture Patterns for LLM-Integrated Applications

Carson Rodrigues, Oysturn Vas

原文摘要

The Model Context Protocol (MCP), introduced by Anthropic in November 2024, defines a standardized interface for connecting large language models (LLMs) to external tools, data sources, and services. Within months of release, hundreds of community-built MCP servers appeared on GitHub, but no software-maintenance literature has yet described how the ecosystem is being structured in production. This industry experience paper catalogues five recurring MCP server architectural patterns observed across an enumerated corpus of fifteen independently developed servers (five production servers from the ANSYR voice AI platform plus ten public servers from the official MCP registry): Resource Gateway, Tool Orchestrator, Stateful Session Server, Proxy Aggregator, and Domain-Specific Adapter. Each pattern is described in the structured form established by Gamma et al.: context, problem, solution, and consequences. We also document four anti-patterns and a set of cross-cutting concerns around authentication, versioning, and observability. Quantitative evaluation contributes two measurements: inter-rater reliability of the taxonomy across two independent LLM raters on 54 held-out servers (Cohen's kappa = 0.76), which also localizes three pattern-boundary ambiguities; transport overhead measured end-to-end on loopback (stdio: 0.01 ms p50; streamable-http: 0.39 ms p50) and modeled for cross-host paths from same-region network baselines (~30 ms p50 baseline plus protocol overhead). Version 2 withdraws the tool-count analysis of version 1, whose numbers could not be traced to production records. Code, corpus, and prompts are released at https://github.com/rodriguescarson/mcp-patterns-icsme2026.

cs.SE · cs.AI

版本更新 · v2arXiv:2611.03597

Structured Visual Landscape: Generating Preferred Representations in Biological and Artificial Neural Networks

Yusong Zhao, Edgar Y. Walker, Hao Wang, Mi Lu

原文摘要

Understanding how neurons respond to visual stimulus is an important question in both deep learning and neuroscience. It has significant implications for enhancing the interpretability of black-box artificial neural networks and for understanding the visual representation in the human brain. We propose a structured visual representations landscape and design an activation-based prior that allows us to effectively regularize the landscape with activations from brain regions or neural network units. Our model Vis-Lens integrates a variational autoencoder and a diffusion model into an image generation framework. It enables the generation of natural, realistic preferred images by directly modifying the activation-regularized latent codes without per-image iterative optimization. We demonstrate the effectiveness of our framework on both artificial neural networks and human visual cortex using functional magnetic resonance imaging (fMRI) data. Our framework improves activation modulation and image realism compared with a recent state-of-the-art method for generating preferred visual representations of these neural targets. Code is available at https://github.com/NeuroAIHub/Vis-Lens.

cs.CV · cs.LG · cs.NE