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

+197新增论文
+151其中 AI 相关
+16已有论文修订
+91新增 AI 导读

arXiv 最近完成扫描:2026-11-22 08:44 · 全站最近入库:2026-11-21 12:19各信源进度 →

统计口径与数据新鲜度

统计 2026-11-21 11:02 — 2026-11-22 11:02(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:2026-11-22 08:44最近发现新增版本:2026-11-21 12:19库内论文最新更新时间:2026-11-20 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。

29,158 已收录论文24,181 AI 相关8,992 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 21,490 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-22 08:42(北京时间)
版本更新 · v2arXiv:2606.15964

PromptShift-CRC: What Drift-Aware Weighting Adds to Online Conformal Risk Control for Foundation Models

Jeffery Opoku, David Banahene

原文摘要

A risk threshold calibrated on old prompts can stop being valid when prompts to a foundation model drift. We study PromptShift-CRC, a rule that weights past prompts by feature similarity, falls back to recent prompts when the current prompt is unlike the past, and updates its level online. We prove that the update controls the long-run violation rate for any weights, so marginal coverage cannot separate weighting schemes. We give a conditional bound at a fixed level whose cost is a weighted total variation distance, bounded by a computable prompt-drift term up to an unknown Lipschitz constant. We propose a comparison protocol: a matched short-window baseline, a validity and sharpness frontier, 30-seed synthetic streams, and real-output streams ordered without using the model's output. On MMLU, TruthfulQA (Qwen2.5-1.5B) and RealToxicityPrompts (GPT-2, replicated with a second generator and classifier), every online method reaches the target. PromptShift-CRC stays within 0.0055 of the Adaptive CRC frontier and avoids the worst window, but DtACI matches or beats it in most settings. The rule performs like a mid-ranked window without having to choose one. Tests fixed in advance to favor the weights show no consistent gain over the frontier. Code, outputs and streams are released.

stat.ML · cs.LG

版本更新 · v2arXiv:2604.07628

Trilinear Compute-in-Memory Architecture for Energy-Efficient Transformer Acceleration

Md Zesun Ahmed Mia, Jiahui Duan, Kai Ni, Abhronil Sengupta

原文摘要

Self-attention in Transformers generates dynamic operands that force conventional Compute-in-Memory (CIM) accelerators into costly non-volatile memory (NVM) reprogramming cycles, degrading throughput and stressing device endurance. Existing solutions either reduce but retain NVM writes through matrix decomposition or sparsity, or move attention computation to digital CMOS at the expense of NVM density. We present TrilinearCIM, a Double-Gate FeFET (DG-FeFET)-based architecture that uses back-gate modulation to realize a three-operand multiply-accumulate primitive for in-memory attention computation without dynamic ferroelectric reprogramming. Evaluated on BERT-base (GLUE) and ViT-base (ImageNet and CIFAR), TrilinearCIM outperforms conventional CIM on seven of nine GLUE tasks while achieving up to 46.6\% energy reduction and 20.4\% latency improvement over conventional FeFET CIM at 37.3\% area overhead. To our knowledge, this is the first architecture to perform complete Transformer attention computation exclusively in NVM cores without runtime reprogramming.

cs.AR · cs.ET · cs.NE

新论文 · v1arXiv:2611.10305

Plug-and-Play Verifier for Decoupling Process Reward in Stabilizing Long-Horizon Agentic Reasoning

Mingwei Xu, Jieyuan Pei, Binghao Tu, Hao Fang

原文摘要

Stabilizing long-horizon reasoning requires intelligent agents to sustain useful progress across successive reasoning steps and tool interactions. Such agents are typically trained with reinforcement learning (RL), where actor-critic RL uses a learnable critic network to estimate step advantages, while critic-free methods such as GRPO eliminate critic models and rely mostly on the group variants of sparse outcome-only rewards. Both reward types leave partially useful failed trajectories either implicit or indistinguishable from entirely unproductive ones, and as reasoning horizons grow, it becomes difficult to maintain consistent reasoning trajectories. In this work, we propose a \textit{plug-and-play verifier} framework for stabilizing long-horizon reasoning through explicit verification of intermediate reasoning artifacts. Specifically, the framework decouples process verification through a unified interface that extracts intermediate reasoning artifacts from agent running trajectories, evaluates them using domain-specific verifiers, and aggregates the verification scores as bounded process reward. Then, we combine our dense verified process rewards and the sparse outcome rewards through a two-stage post-training schedule. We showed that with plug-and-play verifiers, we further stabilize the reasoning steps as last-mile self-reflection and self-consistency. Typical evaluation on HotpotQA using Qwen3.5-4B achieves a superior accuracy (60.18\%), outperforming the standard GRPO baseline (57.73\%). Our results suggest that plug-and-play verifiers provide a modular approach to learn more stable and reliable reasoning behaviors over long horizons.

cs.LG

版本更新 · v3arXiv:2410.02504

Dual Active Learning for Reinforcement Learning from Human Feedback

Pangpang Liu, Chengchun Shi, Will Wei Sun

原文摘要

Aligning large language models (LLMs) with human preferences is critical to recent advances in generative artificial intelligence. Reinforcement learning from human feedback (RLHF) is widely applied to achieve this objective. A key step in RLHF is to learn the reward function from human feedback. However, human feedback is costly and time-consuming, making it essential to collect high-quality conversation data for human teachers to label. Additionally, different human teachers have different levels of expertise. It is thus critical to query the most appropriate teacher for their opinions. In this paper, we use reinforcement learning (RL) to formulate the alignment problem. Motivated by the idea of $D$-optimal design, we first propose a dual active reward learning algorithm for the simultaneous selection of conversations and teachers. Next, we apply pessimistic RL to solve the alignment problem, based on the learned reward estimator. Theoretically, we show that the reward estimator obtained through our proposed adaptive selection strategy achieves minimal generalized variance asymptotically, and prove that the sub-optimality of our pessimistic policy scales as $O(1/\sqrt{T})$ with a given sample budget $T$. Through simulations and experiments on LLMs, including annotator-level preference data and downstream Best-of-$N$ response-selection evaluations, we demonstrate the effectiveness of our algorithm and its superiority over existing baselines.

stat.ML · cs.LG

版本更新 · v4arXiv:2510.02300

Equilibrium Matching: Generative Modeling with Implicit Energy-Based Models

Runqian Wang, Yilun Du

原文摘要

We introduce Equilibrium Matching (EqM), a generative modeling framework built from an equilibrium dynamics perspective. EqM discards the non-equilibrium, time-conditional dynamics in traditional diffusion and flow-based generative models and instead learns the equilibrium gradient of an implicit energy landscape. Through this approach, we can adopt an optimization-based sampling process at inference time, where samples are obtained by gradient descent on the learned landscape with adjustable step sizes, adaptive optimizers, and adaptive compute. Empirically, EqM achieves the best FID among the tested methods, reaching 1.90 on ImageNet 256$\times$256, while remaining competitive on other metrics. Theoretically, we show under standard assumptions that real images are attracting critical points of the learned field and the energy interpretation applies when the field is conservative. Beyond generation, EqM is a flexible framework that naturally handles tasks including partially noised image denoising, OOD detection, and image composition. By replacing time-conditional velocities with a unified equilibrium landscape, EqM offers a tighter bridge between flow and energy-based models and a simple route to optimization-driven inference.

cs.LG · cs.AI · cs.CV

新论文 · v1arXiv:2611.10298

Mobility Information Capacity of Multi-Robot Systems: Safety-Aware Maneuver Design

Weijie Yuan

原文摘要

Multiple robots operating in shared workspaces must adjust their speed, direction, and trajectories to coordinate with one another and respond to changing tasks and disturbances. The available maneuver freedom depends on the dynamics, uncertainty, and safety margins, even for a fixed fleet size and workspace. However, robot count, workspace occupancy, and task throughput do not quantify how much the fleet can vary its motion or how reliably these variations can be observed. We develop a mobility information capacity (MIC) framework that jointly characterizes these two aspects. Mobility information is defined as the mutual information between a joint maneuver sequence and its noisy trajectory observation, and MIC maximizes this quantity under maneuver-resource constraints. For a linear Gaussian maneuver-to-observation model with feedback dynamics, we characterize the finite-horizon information benchmark and derive its stationary long-horizon spectral limit, accounting for heterogeneous mobility gains, colored uncertainty, and temporally coupled maneuver costs. We further map workspace-boundary, inter-robot separation, and actuator chance constraints into coupled maneuver-covariance restrictions and derive conditions under which covariance optimization reduces to resource allocation across mobility modes. Finite joint maneuver codebooks are then constructed through individual risk screening and noise-whitened trajectory separation, with bounds linking sequence recovery error and achieved information to the capacity benchmarks. An aerial-robot simulation study evaluates spectral convergence, safety-aware allocation, finite-codebook identification, and fleet-size scaling. The resulting framework complements occupancy and task-throughput measures by quantifying the maximum observable maneuver freedom supported by a mobility system.

cs.RO · cs.IT · eess.SY

新论文 · v1arXiv:2611.10297

Uncertainty, Not Concealment: A Mechanistic Reappraisal of Sycophancy in Vision-Language Models

Mohamed I. Eldagla, Zeyad H. ElShafei, Ahmed M. Anter

原文摘要

Vision-language models (VLMs) in medical retrieval-augmented generation must be able to say when a retrieved report contradicts the X-ray, so understanding whether they can detect such conflicts is important for safe deployment. The sycophancy mask hypothesis holds that these models internally detect conflicts but refuse to report them, concealing a detected contradiction. We introduce a three-part diagnostic that separates concealment from uncertainty: representation probing, activation steering, and two-way forced choice. Across four instruction-tuned VLMs on chest X-rays, we confirm behavioral suppression of contradiction and find that a linear probe separates matched from mismatched inputs, beyond what a simple CLIP model achieves. Steering the learned conflict direction never elicits contradiction, and forcing a binary choice does not recover one; instead, cross-modal attention to the image drops on mismatched pairs. We conclude that these models fail to contradict because the mismatch registers as hedged, gated uncertainty rather than concealed knowledge, calling for re-grounding and uncertainty surfacing. Code is available at https://github.com/mohamed-eldagla/Uncertainty-Not-Concealment/.

cs.CV · cs.LG

新论文 · v1arXiv:2611.10294

Robust Analysis of First-Price Auctions with Large Supply

Jason Hartline, Anant Shah, Matthew vonAllmen

原文摘要

This paper performs a robust analysis of welfare (i.e., the price of anarchy) in multi-unit first-price auctions with independent private values and unit-demand bidders. We employ a smoothness framework (see \citealp{ST-13}) to analyze the price of anarchy when supply is large. The main barrier to analysis in this framework is a worst-case bid distribution that establishes a lower bound on the price of anarchy of $1- \frac{1}{e}$. We show that such worst-case distributions cannot arise in equilibrium when agents' values satisfy a mild density lower bound condition. This characterization leads to novel asymptotic lower bounds. In particular, the price of anarchy for $ε$-approximate equilibrium as the number of units goes to infinity is at least $1-Θ\left(ε\log \left(\frac{1}ε\right)\right)$. Our analysis relies on a key structural property of equilibrium bid distributions: for a welfare analysis in any multi-unit first-price auction, it is without loss to analyze (approximate) equilibria with no atoms. In addition to this, we establish continuity and monotonicity properties of Bayes-Nash equilibria. We believe these results are of independent interest.

cs.GT

新论文 · v1arXiv:2611.10293

Sparse Predict and Demote: Optimal Scheduling with Minimal Size Advice

Chengyi Liu, Thodoris Lykouris, Junxuan Shen, Adam Wierman

原文摘要

Knowledge of job sizes is crucial for optimizing the mean response time in single-server scheduling. However, predictions of job sizes are typically inexact and may be costly to obtain. Thus, a long-standing research question is: what is the minimal information about job sizes required to obtain near-optimal mean response time? We introduce a novel policy, \emph{Sparse Predict and Demote (SPD)}, and prove that it achieves order-optimal mean response time in heavy traffic while requiring only one-bit job-size predictions for a vanishing fraction of arrivals. SPD achieves heavy-traffic order-optimality even in situations when predictions are (i) inexact up to any fixed multiplicative error bound, and (ii) costly, i.e., the server is used to compute predictions and is thus unavailable to serve jobs while processing prediction queries. Finally, we use numerical experiments to illustrate that SPD maintains near-optimal performance beyond the heavy-traffic regime.

cs.PF · cs.DS · cs.LG

版本更新 · v2arXiv:2610.40959

P-ContextBench: From Personal Fact Recall to Personalized Agent Behavior

Grant Nasir Brown, Abdul Basir Bari, Anmol Tadikonda, Isabel Song

原文摘要

Large language models (LLMs) are increasingly used for personal tasks, yet evaluating their safe use of personal context remains difficult because such contexts are private, fragmented, dynamic, and socially situated. We introduce P-ContextBench, a realistic, verifiable, and scalable benchmark for evaluating personal-context understanding. P-ContextBench constructs population-structured, life-story-grounded digital humans from demographic seeds reflecting the U.S. population. Rather than using explicit, static profiles, P-ContextBench embeds traits, routines, goals, and constraints as implicit evidence across fragmented, evolving synthetic mobile-app records. It evaluates five capabilities: single-source fact recovery, multi-source fact recovery, temporal state tracking, inference of higher-level personal patterns from routine behavior, and safe handling of requests conflicting with persona-specific constraints through compliance, warning, clarification, or refusal. A multi-stage validation pipeline checks persona plausibility, context recoverability, information leakage, evidence preservation, test-case grounding, and human-judge agreement. We instantiate P-ContextBench with 100 personas, 10,088 hidden facts, 18,261 app-trace fragments, and 4,810 verified test cases. The pipeline supports configurable generation across scales and demographic groups. Cost-effective state-of-the-art (SOTA) LLMs relevant to personal agents achieve below 60% accuracy on temporal tracking, higher-level pattern inference, and constraint-aware action-boundary decisions, with fewer than 10% of test cases answered correctly across all three repeated runs. Even a high-performing SOTA model remains inconsistent on these capabilities. P-ContextBench provides measurable targets for improving the accuracy, reliability, and safety of LLM-based personal agents.

cs.AI

新论文 · v1arXiv:2611.10288

A Clinical Evaluation of Conformal Prediction for Chest Radiograph Report Generation

Nikolaos Kontogeorgis, Pascal Krezer, Klaus-Rudolf Kladny, Arne Estler, Christian F. Baumgartner, Lisa Koch, Saif Afat, Michael Muehlebach

原文摘要

While multimodal large language models (MLLMs) have improved automated radiology report generation, they remain prone to hallucinations, necessitating clinically meaningful risk-control methods. Conformal prediction (CP) provides statistical guarantees, yet existing CP approaches rely on uncalibrated automated extractors and fail to account for the clinical relevance of errors. We introduce clinical admissibility criteria (CLARITY), a framework for evaluating radiology reports that explicitly distinguishes clinically consequential errors from minor deviations. To eliminate reliance on automatic judges, we collected radiologist annotations of MLLM-generated reports using CLARITY. By benchmarking CP strategies and automatic evaluation metrics against these expert annotations, we demonstrate that commonly used admissibility proxies systematically fail to detect clinically relevant errors. Our findings reveal that automated admissibility judgments are largely decoupled from true clinical admissibility.

cs.LG

版本更新 · v3arXiv:2603.29501

Target-Aligned Reinforcement Learning

Leonard S. Pleiss, James Harrison, Maximilian Schiffer

原文摘要

Many value-based deep reinforcement learning algorithms rely on target networks - lagged copies of the online network - to stabilize training. While effective, this mechanism introduces a fundamental stability-recency trade-off: slower target updates improve stability but reduce the recency of learning signals, hindering convergence speed. We propose Target-Aligned Reinforcement Learning (TARL), a simple drop-in refinement for existing algorithms that emphasizes transitions for which the target and online network estimates are highly aligned. By focusing updates on well-aligned targets, TARL mitigates the adverse effects of stale target estimates while retaining the stabilizing benefits of target networks. We empirically demonstrate consistent improvements within discrete and continuous control algorithms across various benchmark environments without any hyperparameter tuning, including a 38.18% median peak score gain on Atari-10, while incurring less than a 4% increase in wall-clock time.

cs.LG · cs.AI

版本更新 · v3arXiv:2510.11052

Latent Refinement Decoding: Enhancing Diffusion-Based Language Models by Refining Belief States

Qinglin Zhu, Yizhen Yao, Runcong Zhao, Yanzheng Xiang, Siya Qi, Amrutha Saseendran, Chen Jin, Philip Teare 等 11 位作者

原文摘要

Autoregressive (AR) models remain the dominant paradigm for natural language generation, but their strictly sequential decoding process leads to high inference latency. Recent diffusion-inspired language models, such as LLaDA and Dream, alleviate this issue through parallel generation, yet they still face two key limitations: information loss, since predictive distributions over non-finalised tokens are discarded at each step, and unstable commitment dynamics, where local token decisions are not sufficiently coordinated at the global level. We propose Latent Refinement Decoding (LRD), a two-stage decoding framework consisting of Latent Refinement and a Predictive Feedback Loop. In the first stage, LRD keeps masked positions as distributional mixtures of predicted tokens and the mask embedding, enabling the model to form more globally consistent beliefs before committing tokens. In the second stage, LRD progressively finalises confident tokens while preserving uncertain ones for further iterative refinement, with KL-divergence dynamics serving as a stable criterion for convergence and early stopping. Experiments show that LRD improves performance on both coding tasks, including HumanEval by 6.3 and MBPP by 2.6, and reasoning tasks, including GSM8K by 2.9 and MATH500 by 3.8, while achieving up to 10.6x overall speedup. Moreover, LRD is compatible with single-token decoding, multi-token threshold commitment, and system-level accelerators; when combined with Fast-dLLM, it reaches up to 18.4x speedup over vanilla decoding while further improving accuracy.

cs.CL

新论文 · v1arXiv:2611.10283

Regev's Reduction Reaches Reconstruction

Stephen P. Jordan, Quynh T. Nguyen, Leo Zhou

原文摘要

Regev's reduction uses a quantum Fourier transform to carry out reductions between various problems related to lattices and codes. We apply it to the problem of sampling from Gibbs distributions of certain classical spin glass Hamiltonians, which can equivalently be thought of as cost functions for average-case instances of an optimization problem called $D$-regular max-$k$-LINSAT. This problem becomes hard at low temperature. Here, we use a quantum decoding algorithm called Belief Propagation with Quantum Messages (BPQM) to solve the decoding problem arising from Regev's reduction. Our main contribution is a proof that the minimum temperature reachable using Regev+BPQM is exactly equal to the reconstruction threshold temperature for noisy channels transmitting signals on a tree network. This equality can be understood via Renes' duality between classical and quantum channels, which we use to show that quantum decoding is possible exactly when classical reconstruction on the dual problem is impossible. M{é}zard and Montanari identified this reconstruction temperature with a dynamical glass transition predicted by the one-step replica-symmetry breaking (1RSB) cavity equation. It is conjectured that local samplers such as Metropolis Monte Carlo and Glauber dynamics cannot equilibrate efficiently below this reconstruction temperature but may do so above it; our large numerical experiments show close agreement with this prediction across hundreds of different problem parameters. Therefore, our results rigorously show that Regev+BPQM reaches a temperature for which, to our knowledge, there is no matching classical guarantee, and brings Regev+BPQM to the conjectured limit of efficient Gibbs sampling by classical local dynamics.

quant-ph · cond-mat.dis-nn · cs.DS

新论文 · v1arXiv:2611.10280

Who Taught the Robot That? Removing One User's Corrections from a Robot Policy That Learns Online

Eshika Pathak, Leela Krishna

原文摘要

A robot that keeps learning after deployment is corrected by several people in turn, and one of them may later ask to be removed. We show that whether the robot can forget that person depends on how people correct it. The goal of removal is the robot that would exist if the person had never been there, which we compute exactly in two simulated tasks with scripted users. When users correct the robot by taking over control, an ordered log of training updates is enough: retraining on the remaining sessions in their original order gives a robot as close to the goal as training noise allows. Deleting the person's data and retraining in one pass does not, even when the other users' corrections do not depend on the removed person, because it loses the order of training. When users only nudge the robot's action, the other users' corrections carry the removed person's influence, and no method we tested recovers the goal robot. Continuing to train without the person removes most of their habits, but not what their data in the training buffer keep alive. The log can identify one clearly harmful user, but not rank users by influence better than counting their corrections.

cs.RO · cs.HC

版本更新 · v3arXiv:2511.20867

E-GEO: A Testbed for Generative Engine Optimization in E-Commerce

Puneet S. Bagga, Vivek F. Farias, Tamar Korkotashvili, Tianyi Peng, Yuhang Wu

原文摘要

With the rise of large language models (LLMs), generative engines have become powerful alternatives to traditional search, creating the need for generative engine optimization (GEO)---improving content visibility and relevance for generative engines. Despite its growing importance, current GEO practices are largely ad hoc, and their impacts remain understudied. In e-commerce, for instance, conversational shopping agents now guide consumers to relevant products, yet it is unclear if GEO can affect how products surface to users. We address this gap by introducing E-GEO, which contains 13,746 realistic, multi-sentence consumer product queries, each paired with 10 retrieved Amazon listings, capturing rich intent, constraints, preferences, and shopping contexts that existing datasets miss. Using this dataset, we conduct a large-scale empirical study of e-commerce GEO across five representative generative engines, seven LLM rewriters, and fifteen rewriting heuristics. We further formulate GEO as an optimization problem and develop a prompt meta-optimization algorithm that significantly improves over heuristic baselines. Notably, prompts independently optimized from different heuristics converge to a stable pattern, suggesting a shared, transferable GEO strategy. Finally, we red-team the GEO system through both static and optimization-based attacks and show that, under a simple in-prompt defense in our setting, gains from GEO reflect genuine content improvement, anchoring GEO as a substantive optimization problem.

cs.IR

新论文 · v1arXiv:2611.10277

Adversarial Evaluation of Large Language Models for Patient-Friendly Radiology Report Summarization

Magdalini Paschali, Asad Aali, Arogya Koirala, Andrew Johnston, Katherine McNamara, Richard Gaus, Roxana Daneshjou, David B. Larson 等 9 位作者

原文摘要

Purpose: To develop and validate a scalable adversarial evaluation framework for assessing the robustness of large language models (LLMs) that generate patient-friendly summaries of radiology reports. Methods: We introduced a clinically grounded adversarial framework that uses controlled term substitutions and impression removals across intensity levels. Using a radiologist-validated adversarial dictionary, we generated adversarial variants of CT and MRI reports. Six LLMs summarized original and adversarial impressions using 2 prompt styles, yielding 18,000 evaluation pairs. Robustness was evaluated with an automated LLM-as-a-judge framework scoring factual faithfulness, sensitivity to adversarial edits, certainty fidelity, and hallucination rate. Failure mode analysis quantified vulnerabilities in incorporating clinical modifiers. Agreement with two radiologists was assessed on 105 cases. Results: GPT-5-mini achieved the highest factual faithfulness and certainty fidelity under both adversarial edits. Under term substitution, hallucinations occurred in 28.1-30.1% of GPT-5-mini summaries, compared with 35.3-36.4% for Claude Sonnet 4.5 and 41.4-41.9% for MedGemma. Impression-removal edits were handled more robustly than term substitutions. Even for the highest-scoring summarizer, substitutions of anatomical terms were not reflected in 23.7% of summaries, diagnostic certainty in 22.3%, and descriptor terms (e.g., "substantial") in 21.2%. Judge-radiologist agreement (macro-F1: 0.55-0.56) was comparable to inter-radiologist agreement (0.53). Conclusions: Adversarial evaluation exposes clinically meaningful failure modes, including missed anatomical modifiers, altered diagnostic certainty, and omitted severity descriptors.

cs.AI

版本更新 · v2arXiv:2504.19152

The Trichotomy of Regular Property Testing

Gabriel Bathie, Nathanaël Fijalkow, Corto Mascle

原文摘要

Property testing is concerned with the design of algorithms making a sublinear number of queries to distinguish whether the input satisfies a given property or is far from having this property. A seminal paper of Alon, Krivelevich, Newman, and Szegedy in 2001 introduced property testing of formal languages: the goal is to determine whether an input word belongs to a given language, or is far from any word in that language. They constructed the first property testing algorithm for the class of all regular languages. This opened a line of work with improved complexity results and applications to streaming algorithms. In this work, we show a trichotomy result: the class of regular languages can be divided into three classes, each associated with an optimal query complexity. Our analysis, which reasons directly and combinatorially on automata, yields effective characterizations of all three classes using so-called minimal blocking sequences.

cs.DS · cs.CC

版本更新 · v2arXiv:2602.03461

Soft-Radial Projection for Constrained End-to-End Learning

Philipp J. Schneider, Daniel Kuhn

原文摘要

Constrained end-to-end learning trains neural networks whose outputs must satisfy feasibility constraints, such as resource, budget, or operational limits. While projection and optimization layers guarantee constraint satisfaction, boundary-based projections can introduce unfavorable optimization geometry: regions of the unconstrained output space are mapped to the same active face of the feasible set. This makes the resulting map locally rank-deficient, suppressing gradients in constrained directions and potentially impairing optimization - a bottleneck known as gradient saturation. We propose Soft-Radial Projection (SRP), a differentiable reparameterization layer that maps unconstrained network outputs into the relative interior of a convex feasible set by smoothly rescaling rays from a strictly feasible anchor point. Unlike hard radial or orthogonal projections, the resulting map is one-to-one and has a full-rank Jacobian almost everywhere, while guaranteeing strict feasibility. We prove that constrained networks equipped with SRP retain universal approximation guarantees. For polytopes and balls, the layer admits a closed-form forward pass, while for general convex sets, the ray-boundary intersection reduces to a one-dimensional search. Experiments on decision-focused learning benchmarks demonstrate strong solution quality at low computational cost, with particularly favorable results in portfolio optimization and substantially lower inference latency than differentiable optimization layers in the evaluated settings.

cs.LG · math.OC · q-fin.CP · stat.ML

版本更新 · v2arXiv:2603.26994

ImmSET: Sequence-Based Predictor of TCR-pMHC Specificity at Scale

Marco Garcia Noceda, Matthew T Noakes, Andrew FigPope, Daniel E Mattox, Bryan Howie, Harlan Robins

原文摘要

T cells are a critical component of the adaptive immune system, playing a role in infectious disease, autoimmunity, and cancer. T cell function is mediated by the T cell receptor (TCR) protein, a highly diverse receptor targeting specific peptides presented by the major histocompatibility complex (pMHCs). Predicting the specificity of TCRs for their cognate pMHCs is central to understanding adaptive immunity and enabling personalized therapies. However, accurate prediction of this protein-protein interaction remains challenging due to the extreme diversity of both TCRs and pMHCs. Here, we present ImmSET (Immune Synapse Encoding Transformer), a novel sequence-based architecture designed to model interactions among sets of variable-length biological sequences. We train this model across a range of dataset sizes and compositions and study the resulting models' generalization to pMHC targets. We describe a failure mode in prior sequence-based approaches that inflates previously reported performance on this task and show that ImmSET remains robust under stricter evaluation. In systematically testing the scaling behavior of ImmSET with training data, we show that performance scales consistently with data volume across multiple data types and compares favorably with the pre-trained protein language model ESM2 fine-tuned on the same datasets. Finally, we demonstrate that ImmSET can outperform AlphaFold2 and AlphaFold3-based pipelines on TCR-pMHC specificity prediction when provided sufficient training data. This work establishes ImmSET as a scalable modeling paradigm for multi-sequence interaction problems, demonstrated in the TCR-pMHC setting but generalizable to other biological domains where high-throughput sequence-driven reasoning complements structure prediction and experimental mapping.

cs.LG · q-bio.QM

版本更新 · v3arXiv:2605.25230

Boosting Inference with Guided Reasoning: Stochastic Exploration for Recursive Models

Andrew Corbett, Archit Sood, Anna Tzatzopoulou, Sai-Aakash Ramesh, Tim Dodwell

原文摘要

Recent work on recursive architectures has shown that tiny neural networks can be surprisingly powerful on structured reasoning tasks. The trick is to model reasoning trajectories with a latent dynamical system. We argue that the inference-time behaviour of these architectures is best understood as approximate inference over latent reasoning trajectories, with deterministic recursion as the one-particle, zero-noise limit. We make this view operational through guided reasoning: stochastic perturbations of the reasoning dynamics propose neighbouring trajectories, and the model's existing early-stopping head reweights them online. The framework yields three diagnostics computable from inference traces: local stability, guide alignment, with a label-free necessary condition, and cloud-token entropy. These predict whether the procedure will help and which of its outputs to trust. On Sudoku-Extreme it lifts exact-solve accuracy from $85.9\%$ to $98.0\%$ without retraining, and on PPBench, a four-type pencil-puzzle benchmark, from $62.5\%$ to $77.9\%$ with no tuning; on Maze-Hard the diagnostics flag a misaligned guide, as validation performance later confirms. The same machinery thus characterises both when recursive reasoning has room to improve at the trajectory level and when the model's internal guide can recover said improvements. Code is available at `https://github.com/digiLab-ai/bigr`.

cs.AI

版本更新 · v2arXiv:2605.07972

It Just Takes Two: Scaling Amortized Inference to Large Sets

Antoine Wehenkel, Michael Kagan, Lukas Heinrich, Chris Pollard

原文摘要

Neural posterior estimation has emerged as a powerful tool for amortized inference, with growing adoption across scientific and applied domains. In many of these applications, the conditioning variable is a set of observations whose elements depend not only on the target but also on unknown factors shared across the set. Optimal inference therefore requires treating the set jointly, which in turn requires training the estimator at the deployment set size -- a regime where memory and compute quickly become prohibitive. We introduce a simple, theoretically grounded strategy that decouples representation learning from posterior modeling. Our method trains a mean-pool Deep Set on sets of size at most two, producing an encoder that generalizes to arbitrary set sizes. The inference head is then finetuned on pre-aggregated embeddings, making training cost essentially independent of the deployment set size N. Across scalar, image, multi-view 3D, molecular, and high-dimensional conditional generation benchmarks with N in the thousands, our approach matches or outperforms standard baselines at a fraction of the compute.

cs.LG · cs.AI · hep-ex · hep-ph · stat.ML

新论文 · v1arXiv:2611.10269

Testing CCZ-equivalence to permutations via projections of Walsh zero spaces, with applications to APN functions

Emanuele Bellini

原文摘要

Whether almost perfect nonlinear (APN) permutations exist in even dimension $n\ge 8$ is a long-standing open question, the big APN problem. The only example known in even dimension, on six bits, was found from a quadratic APN function that is not itself a permutation, by Carlet--Charpin--Zinoviev (CCZ) equivalence. It is therefore natural to test APN functions for CCZ-equivalence to a permutation. This amounts to finding two complementary $n$-dimensional subspaces among the zeros of the Walsh transform, which lie in $\mathbb{F}_2^{2n}$. A known necessary condition only uses their projections to $\mathbb{F}_2^n$ and cannot decide every case. We give an exact test for functions whose Walsh supports are affine, including all quadratic functions. It reconstructs the subspaces from their projections, through searches for subspaces of $\mathbb{F}_2^n$ and for constrained bilinear maps. On the benchmarked functions in dimension twelve, it lists them $39$ to $55$ times faster than the generic search of the commonly used sboxU library. We then study the functions $x^3+a^{-1}\mathrm{Tr}(a^3x^9)$, which Göloğlu and Pavlů proved not to be CCZ-equivalent to permutations when $4$ divides $n$. With their result, our computations show that this holds for every even $n\le 24$. In dimension $18$ the necessary condition is satisfied, and the new test decides the case. We conjecture this for every even dimension. We also compile $25$ classes of quadratic APN functions in dimension ten from published instances, with equivalence certificates. No function in these classes or in their switching closure is CCZ-equivalent to a permutation. All code and data are public.

cs.CR · cs.IT

版本更新 · v2arXiv:2610.33133

PINND: Physics Informed Neural Networks for Solving the Drift-Diffusion Equations

Luke Ragan, Matteo G. C. Alasio, Enrico Bellotti, Luca Dal Negro

原文摘要

The drift-diffusion equations are the fundamental building block for the modeling, design, and optimization of modern semiconductor devices. Solutions of the drift-diffusion equations rarely give way to analytical forms, and exhibit strong nonlinearity, with the solutions ranging over several orders of magnitude. These attributes have led to research into numerical solvers, and while many commercial Technology Computer Aided Design (TCAD) packages now exist, they may require long simulation times, exhibit sensitivity to initial guesses, and rely on mesh generation. Further, these solvers are not well suited for inverse problems where the design parameters of a device are retrieved from its desired functional characteristics. To address all these challenges, this work demonstrates the usage of Physics Informed Neural Networks for the Drift-diffusion (PINND) equations with errors comparable to traditional simulation software using advanced training techniques to solve problems beyond the well-behaved equilibrium regime. For this study, we specifically choose an abrupt PN-junction in both one and two dimensions that features extreme gradients of the carrier concentrations at the Ohmic contacts and an abrupt discontinuity in the doping profile. Using the proposed PINN architecture, we are able to retrieve accurate results for all observables under all bias conditions. Additionally, we apply the PINN solver towards the inverse problem of retrieving doping profiles directly from desired electric field distributions or electrostatic potentials under non-zero applied biases. Our work establishes a powerful PINNs-based approach for semiconductor device modeling with the ability to solve inverse problems under non-equilibrium conditions for microelectronics research and discovery.

cs.LG

新论文 · v1arXiv:2611.10266

I play of the cello: Evaluating Constructional Bleeding in Multilingual LLMs with CALQUE

Toshiki Nakai, Yu-Yin Hsu, Marie-Catherine de Marneffe, Joakim Nivre, Jakob Prange, Nathan Schneider, Wesley Scivetti, Tiago Timponi Torrent 等 9 位作者

原文摘要

Languages use different morphosyntactic strategies to express the same meaning. Speakers of multiple languages are able to apply strategies from one language to another, either by mistake or intentionally, to produce a literal translation, or calque. Considering that large language models (LLMs) have been shown to generalize crosslingually, despite being predominantly trained on English, we investigate whether they overgeneralize constructional strategies from English to other languages. To test this, we create a dataset of 5583 sentence pairs across seven languages, matching literal translations of strategies from English into other languages, and literal translations of the other languages' strategies into English. We then quantify constructional bleeding by measuring how often strategies from English are assigned a higher probability than native strategies. We find that models often fail to assign higher probabilities to native strategies and tend to prefer English-derived strategies even when evaluating sentences in other languages. Our findings highlight the limitations of the current English-dominant training paradigm.

cs.CL

新论文 · v1arXiv:2611.10265

Do large language models rank moral values? A transitivity test of moral competence

Katelyn A. Wasson, Kiwon Song, James M. Jennings, Brittney Currie, Konstantina Sokratous, Clintin P. Stober

原文摘要

LLMs are increasingly used to support decisions that require moral consideration. While LLMs perform well on morality benchmarks from psychology and philosophy, producing appropriate outputs on these benchmarks (moral performance) does not establish that an LLM weighs competing moral values at stake in a decision (moral competence). We test a minimal, necessary condition for such weighing: that an ordinal ranking can represent LLM moral choices. Such a ranking requires transitivity. To assess transitivity, we presented LLMs with pairwise trade-offs among moral value violations in vignette choice sets and extracted resulting choice probabilities from first-token distributions. Tested against three stochastic transitive choice models, most choice sets admitted weak or unstable orderings. Aggregating an LLM's responses, a common practice in LLM moral evaluation, produced transitive structure that matched no individual choice set. Evaluations of AI morality should establish that choices are structured before asking whether they are aligned.

cs.AI · cs.CL · cs.CY

版本更新 · v2arXiv:2607.22862

Sensitivity Analysis of the Transformation Using a Parameterized CEC 2017 Benchmark

Grzegorz Sroka, Sławomir T. Wierzchoń

原文摘要

CEC 2017 applies bias, shift and rotation simultaneously, obscuring their individual effects on algorithm behavior. We introduce a parameterized implementation controlling them independently while preserving the original functions and transformation data. It diagnoses hybrid Marine Predators Algorithm (hMPA), whose predicted-candidate mechanism relies on numerical objective values and coordinate-wise reconstruction. Deep Statistical Comparison (DSC) and extended DSC (eDSC) are adapted from algorithm-level comparison to configuration-level diagnosis. To our knowledge, this is the first exhaustive DSC and eDSC analysis of all eight bias-shift-rotation configurations of a parameterized CEC 2017 benchmark. We examine all 56 three-configuration subsets and comparisons with the untransformed control, using 29 functions, dimensions 10, 30, 50 and 100, 30 independent runs and a fixed budget of 10,000 objective-function evaluations per dimension. DSC and eDSC show statistically significant configuration differences at every dimension, but rankings vary across functions, precluding a consistent ordering. Most individual effect-size confidence intervals include zero. Among those excluding zero, comparisons favor MPA more often at dimension 10 and hMPA at dimensions 50 and 100. Shift generally worsens objective-value outcomes. Isolated rotation causes no systematic deterioration; isolated bias barely changes final-solution distributions. Combined shift and rotation differ most consistently from the control. Convergence plots show function- and dimension-dependent separation, plateaus and late improvement. Numerical findings are limited to hMPA, but the framework can be extended to other continuous optimizers and adapted to related CEC benchmarks, including CEC 2024 and CEC 2025.

cs.NE

新论文 · v1arXiv:2611.10264

Look Only Where You Plan for Real-Time End-to-End Urban Driving

Haohua Que, Tianle Zhu, Qianyi Wu, Zhipeng Bao, Handong Yao

原文摘要

Dense bird's-eye-view (BEV) construction accounts for 60% of the parameters in a strong recent driving planner. This paper presents LOWP, a planner that looks only where it plans: two sets of sixteen learnable queries read the cameras through the known calibration, one by sampling image features at projected 3D anchors and one by attending to features labeled with their viewing rays, so no dense BEV is built and no object or map is supervised. Evaluations on open-loop nuScenes, pseudo-simulated NAVSIM and closed-loop Bench2Drive show competitive planning with a compact representation. On nuScenes, average L2 changes from 0.302 to 0.296 m with 53% of the dense counterpart's parameters and 1.7 times its speed on the same GPU; the anchor-based reader performs better on turns than the global reader. On NAVSIM, scoring the planner's own trajectories alongside a vocabulary gives 85.9 PDMS without retraining. The closed-loop control configuration scores 65.5 on 218 routes, with a brake-induced standstill failure whose release rule has an inconclusive aggregate benefit. We distinguish measured latency from hardware-transferred estimates, which do not establish a common-hardware ranking.

cs.RO

版本更新 · v2arXiv:2605.06660

Verifier-Backed Hard Problem Generation for Mathematical Reasoning

Yuhang Lai, Jiazhan Feng, Yee Whye Teh, Ning Miao

原文摘要

Large Language Models (LLMs) demonstrate strong capability in solving scientific and mathematical problems, yet they struggle to produce valid and challenging novel problems -- an essential component for advancing LLM training and enabling autonomous scientific research. Existing problem generation approaches either depend on expensive human expert involvement or adopt naive self-play paradigms, which frequently yield invalid problems due to reward hacking. This work introduces VHG, a verifier-backed hard problem generation framework built upon three-party self-play. By integrating an independent verifier into the conventional setter-solver duality, our design constrains the setter's reward to be jointly determined by problem validity (evaluated by the verifier) and difficulty (assessed by the solver). We instantiate two verifier variants: a Hard symbolic verifier and a Soft LLM-based verifier, with evaluations conducted on indefinite integral tasks and general mathematical reasoning tasks. Experimental results show that VHG substantially outperforms all baseline methods by a clear margin. Our code is available at https://github.com/Miaow-Lab/VHG.

cs.LG · cs.AI · cs.CL

新论文 · v1arXiv:2611.10262

Domain Generalization for Style-Debiased Object Detection via Label-Consistent Training across Multiple Styles

Chaoran Zhang, Shirshak Bhusal, Zhilin Gao, Qiang Xue, Haoting Shen

原文摘要

Domain generalization (DG) for object detection aims to improve model performance on unseen target domains. Existing detectors, however, suffer substantial degradation on visually distinct unseen domains, as models confined to limited training styles tend to overfit superficial statistical cues rather than object-level semantics. Motivated by human perception, which attends primarily to semantically relevant features while ignoring irrelevant style cues, we propose Origin-Anchored Multi-Style Consistency (OAMSC), a source-only framework that couples multi-style data synthesis with cross-style consistency learning to suppress style-sensitive shortcut learning. For data preparation, we leverage off-the-shelf diffusion models to expand the source datasets into 20 distinct styles, with all generated images manually verified to ensure that object instances and annotations remain consistent across style variants. During training, we impose Origin-anchored consistency constraints at scene and object levels to reinforce the learning of domain-stable features while suppressing the influence of style-sensitive cues. Extensive experiments and ablation studies on multiple challenging DG benchmarks demonstrate the consistent effectiveness and strong generalization performance of OAMSC across diverse domain shifts. Notably, our approach applies to both one-stage and two-stage detectors, while its data-centric design is complementary in principle to other DG methods. The training code and two style-transfer datasets constructed for the experiments, comprising approximately 180K images in total, are publicly available on GitHub.

cs.CV