语言大模型 / LLM - arXivDaily 专题

2606.18673 2026-06-18 cs.CR 新提交 91%

Understanding and Mitigating Prompt Leaking Attacks in Real-World LLM-Based Applications

理解并缓解真实世界基于LLM的应用中的提示泄露攻击

Yong Yang, Chong Fu, Tong Zhang, Rui Zeng, Qingming Li, Tianyu Du, Zonghui Wang, Shouling Ji, Wenzhi Chen

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结本研究系统测量了1200个真实世界基于LLM的应用，发现超过80%会泄露系统提示，并提出了基于注意力漂移分析的AREA防御方法，在保持可用性的同时有效防止泄露。

Comments Accepted at ACM CCS 2026

2606.17309 2026-06-17 cs.RO 新提交 91%

Abstention-Aware Personalized Object Rearrangement via Uncertainty-Guided LLM Assistance

基于不确定性引导的LLM辅助的弃权感知个性化物体重排

Sam Collin, Ali Ayub

机构 * Concordia University（康考迪亚大学）

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结提出APOLLO框架，结合轻量级个性化嵌入模型与选择性大语言模型辅助，通过不确定性估计在模糊决策时调用LLM，实现高效、隐私保护的弃权感知物体重排。

Comments Accepted at the 2026 IEEE 35th International Conference on Robot and Human Interactive Communication (RO-MAN 2026)

2606.15473 2026-06-16 q-fin.RM 新提交 91%

Belief at Risk: Quantifying Agentic AI Model Risk with LLM-Inferred Bayesian State Filters

风险中的信念：利用LLM推断的贝叶斯状态滤波器量化智能体AI模型风险

Matthew Francis Dixon

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结提出将智能体AI系统建模为部分可观测马尔可夫决策过程，利用LLM作为语义观测模型，结合贝叶斯滤波器量化不确定性，并计算风险度量，为金融等受监管环境中的智能体AI验证提供严格基础。

Comments 15 pages, 3 figures

2606.10864 2026-06-10 eess.AS 新提交 91%

Phoneme-First Prediction for LLM-Based Speech Recognition

基于LLM的语音识别的音素优先预测

Jakob Poncelet, Hugo Van hamme

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结提出在LLM中集成音素预测步骤，先预测音素再生成转录，以提升低资源场景下的语音识别准确性和可解释性。

Comments Accepted at EUSIPCO 2026

2605.29378 2026-05-29 cs.RO 91%

Decentralized LLM-Driven Coordination of Acoustic Robots for Contactless Object Manipulation

去中心化LLM驱动的声学机器人协调用于非接触式物体操控

Yingying Wang, Narsimlu Kemsaram, Sriram Subramanian

机构 * Department of Computer Science, University College London（计算机科学系，伦敦大学学院）； Department of Artificial Intelligence, University of Malaya（人工智能系，马来大学）

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结提出一种去中心化框架，利用Whisper语音识别和LLM语义解析将自然语言指令转换为多机器人任务计划，实现声学机器人的非接触式物体操控，实验验证了顺序、并行和同步协作任务的有效性。

Comments This paper has been accepted for publication in the Proceedings of the 2026 IEEE 22nd International Conference on Automation Science and Engineering (CASE 2026), August 17-21, 2026, Shenyang, China

2605.04704 2026-05-08 cs.AR cs.SE 91%

UVMarvel: an Automated LLM-aided UVM Machine for Subsystem-level RTL Verification

UVMarvel：一种自动化LLM辅助的UVM机 for 子系统级RTL验证

Junhao Ye, Dingrong Pan, Hanyuan Liu, Yuchen Hu, Jie Zhou, Ke Xu, Xinwei Fang, Xi Wang, Nan Guan, Zhe Jiang

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 UVMarvel利用LLM自动生成子系统级RTL验证的UVM测试平台，通过中间表示和总线协议库实现协议正确性，减少验证时间至4.5小时，代码覆盖率达95.65%。

Comments This paper has been accepted by DAC 2026 and will appear in the proceedings

2604.17097 2026-04-21 cs.AR 91%

From Natural Language to Silicon: The Representation Bottleneck in LLM Hardware Design

从自然语言到硅基：LLM硬件设计中的表示瓶颈

Weimin Fu, Zeng Wang, Minghao Shao, Johann Knechtel, Ozgur Sinanoglu, Ramesh Karri, Muhammad Shafique, Xiaolong Guo

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结研究揭示LLM硬件设计中表示选择而非模型选择是决定最终成功率的关键，通过评估三种前沿LLM在六种不同IR上的表现，发现简单偏见使资源受限的FPGA实现更高效。

2505.18351 2026-04-21 cs.MA cs.CY cs.DB 91%

Persona Alchemy: Designing, Evaluating, and Implementing Psychologically-Grounded LLM Agents for Diverse Stakeholder Representation

人格炼金术：为多样化利益相关者代表设计、评估和实现心理基础的LLM代理

Sola Kim, Dongjune Chang, Jieshu Wang

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结本文提出基于社会认知理论的LLM代理设计框架，通过四个个人因素、六个可量化构念和图数据库架构，提升LLM代理在多样化利益相关者代表中的心理一致性与可解释性。

Comments Accepted at ICLR 2026 Algorithmic Fairness Across Alignment Procedures and Agentic Systems (AFAA) Workshop

2407.04681 2024-07-08 cs.CV cs.AI cs.CL cs.LG 91%

Rethinking Visual Prompting for Multimodal Large Language Models with External Knowledge

Yuanze Lin, Yunsheng Li, Dongdong Chen, Weijian Xu, Ronald Clark, Philip Torr, Lu Yuan

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；prompting(title)；分类 cs.CL、cs.AI、cs.LG

2303.06573 2023-10-23 cs.IR 91%

Large Language Models Know Your Contextual Search Intent: A Prompting Framework for Conversational Search

Kelong Mao, Zhicheng Dou, Fengran Mo, Jiewen Hou, Haonan Chen, Hongjin Qian

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；prompting(title,abstract)

Comments Accepted to Findings of EMNLP 2023. Code: https://github.com/kyriemao/LLM4CS/

2512.13374 2026-08-13 cs.AI 版本更新 91%

Behavior and Representation in Open-Weight Large Language Models for Combinatorial Optimization: From Feature Extraction to Algorithm Selection

大语言模型在组合优化中的行为与表示：从特征提取到算法选择

Francesca Da Ros, Luca Di Gaspero, Kevin Roitero

机构 * University of Udine（乌迪大学）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,abstract_cn)；prompting(abstract)

AI总结本研究探讨了大语言模型在组合优化中的内部表示能力，通过直接查询和探测分析，评估其提取特征和预测最佳求解器的效果。

Comments Accepted for publication in Computers & Operations Research. Code and data are available on Zenodo: https://doi.org/10.5281/zenodo.21891840

2606.29545 2026-06-30 cs.CL 91%

AURORA: Asymmetry and Update-Induced Rotation for Robust Hallucination Detection in Large Language Models

AURORA：用于大型语言模型中鲁棒幻觉检测的不对称性与更新诱导旋转

Zishuai Zhang, Hainan Zhang, Zhiming Zheng

机构 * School of Artificial Intelligence, Beihang University, China（北京航空航天大学人工智能学院）； Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing, Beihang University, China（北京航空航天大学未来区块链与隐私计算先进创新中心）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(summary_cn,abstract_cn)；分类 cs.CL

AI总结提出AURORA框架，利用权重梯度动态（不对称性和旋转比）检测LLM幻觉，跨模型和数据集表现鲁棒。

2604.11087 2026-05-12 cs.LG 91%

CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models

CausalGaze: 通过反事实图干预揭示大语言模型的幻觉

Linggang Kong, Lei Wu, Yunlong Zhang, Xiaofeng Zhong, Zhen Wang, Yongjie Wang, Yao Pan

机构 * College of Electronic Engineering, National University of Defense Technology（国防科技大学电子工程学院）； Anhui Province Key Laboratory of Cyberspace Security Situation Awareness and Evaluation（安徽省网络空间安全态势感知与评估重点实验室）； Institute of Computer Application, China Academy of Engineering Physics（中国工程物理研究院计算机应用研究所）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(summary_cn,abstract_cn)；分类 cs.LG

AI总结 CausalGaze通过反事实图干预揭示大语言模型的幻觉，利用结构因果模型提升模型可解释性，在四个数据集和三个常用LLM上实验表明其有效性，尤其在TruthfulQA数据集上比现有方法提升3.3%的AUROC。

Comments Accepted as ACL2026 Findings

2604.13331 2026-05-05 cs.LG 91%

Text-Attributed Knowledge Graph Enrichment with Large Language Models for Medical Concept Representation

基于大语言模型的文本属性知识图谱增强用于医学概念表示

Mohsen Nayebi Kerdabadi, Arya Hadizadeh Moghaddam, Chen Chen, Dongjie Wang, Zijun Yao

机构 * University of Kansas, USA（堪萨斯大学）

专题命中知识编辑与模型理解：LLM(summary_cn,abstract)；large language model(title)；language model(title)；prompting(abstract)

AI总结本文提出MedCo框架，通过结合统计关联和LLM提示构建医学知识图谱，并利用LLM生成节点描述和边解释，最终融合文本语义与图结构生成统一的概念嵌入，提升临床预测性能。

Comments This paper has been accepted at ACL 2026 main conference

2604.14463 2026-04-17 cs.CL 91%

Psychological Steering of Large Language Models

对大语言模型的心理操控

Leonardo Blas, Robin Jia, Emilio Ferrara

机构 * Department of Computer Science, University of Southern California（南加州大学计算机科学系）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；prompting(abstract,abstract_cn)；LLM(abstract_cn)

AI总结本文提出一种心理操控框架，通过在语义校准的单元中进行无界、流畅性约束的扫描，改进大语言模型的干预方法，在14个模型中表现出色，优于传统方法。

Comments 66 pages, 60 images

2311.03533 2023-11-08 cs.CL 91%

Quantifying Uncertainty in Natural Language Explanations of Large Language Models

Sree Harsha Tanneru, Chirag Agarwal, Himabindu Lakkaraju

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；foundation model(abstract)

2606.21195 2026-06-23 cs.CL cs.AI 新提交 91%

Beyond Hooking Onto the World: Referential Profiles and the Numerical Structure of LLM Grounding

超越钩连世界：指称轮廓与LLM基础的数字结构

Joo Yull Rhee

机构 * Sungkyunkwan University（成均馆大学）

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

AI总结本文重新审视大语言模型的基础问题，提出指称是基于轮廓、上下文敏感、受情感影响且受规范约束的，并通过优化将语言痕迹参数化为数字结构，支持LLM拥有衍生性、语言中介的指称形式。

Comments 29 pages, no figures

2605.23035 2026-05-25 cs.CL cs.AI q-bio.NC 91%

Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography

稀疏自编码器将大脑-LLM对齐映射到皮层语义拓扑

Dongxin Guo, Jikun Wu, Siu Ming Yiu

机构 * The University of Hong Kong（香港大学）； Stellaris AI Limited（Stellaris AI有限公司）

专题命中知识编辑与模型理解：LLM(title,title_cn)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

AI总结通过稀疏自编码器分解LLM中间层特征，发现语义特征主导大脑预测表现，并验证了皮层语义拓扑的细粒度对齐。

Comments Accepted at CoNLL 2026. 20 pages (9 main + 1 limitations/acknowledgments + 3 references + 7 appendix), 5 figures, 20 tables

2505.06330 2025-08-05 cs.LG cs.AI eess.SP 91%

Prompting Large Language Models for Training-Free Non-Intrusive Load Monitoring

Junyu Xue, Xudong Wang, Xiaoling He, Shicheng Liu, Yi Wang, Guoming Tang

机构 * Southern University of Science ； Technology \& Peng Cheng Laboratory ； The Chinese University of Hong Kong, Shenzhen ； The Hong Kong University of Science ； Tsinghua University

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；prompting(title)；分类 cs.AI、cs.LG

2411.16189 2024-11-26 cs.AI cs.CL cs.MA 91%

Enhancing Multi-Agent Consensus through Third-Party LLM Integration: Analyzing Uncertainty and Mitigating Hallucinations in Large Language Models

Zhihua Duan, Jialin Wang

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(title)；分类 cs.CL、cs.AI

2408.04023 2024-08-09 cs.CL cs.AI 91%

Improving Large Language Model (LLM) fidelity through context-aware grounding: A systematic approach to reliability and veracity

Wrick Talukdar, Anjanava Biswas

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(title)；分类 cs.CL、cs.AI

Comments 14 pages

Journal ref World Journal of Advanced Engineering Technology and Sciences, 2023, 10(2), 283-296

2608.11596 2026-08-13 eess.SP 新提交 91%

Small Language Model enabled Autonomous agent for Language-Conditioned Cognitive Radar

基于小型语言模型的语言条件认知雷达自主智能体

Minhaj Uddin Ahmad, Zakia Zaman, Shunqiao Sun, Mizanur Rahman

专题命中知识编辑与模型理解：language model(title,abstract)；small language model(title,abstract)；SLM(abstract,abstract_cn)；prompting(abstract)

AI总结本文提出一种小型语言模型驱动的自主智能体框架，作为语言条件认知雷达的智能控制器，经实验验证可在多种雷达场景下完成算法选择，且雷达特定提示与基于物理的工具执行是可靠决策的必要条件。

Comments Accepted at MLSP 2026, ATL, USA

2510.23636 2026-07-08 cs.LG cs.AI cs.CL 版本更新 91%

LLM4Delay: Flight Delay Prediction via Cross-Modality Adaptation of Large Language Models and Aircraft Trajectory Representation

LLM4Delay：通过大语言模型与飞机轨迹表示的跨模态适应进行航班延误预测

Thaweerath Phisannupawong, Joshua Julian Damanik, Han-Lim Choi

机构 * Korea Advanced Institute of Science and Technology（韩国科学技术院）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,abstract_cn)；分类 cs.CL、cs.AI、cs.LG

AI总结本文提出LLM4Delay框架，通过整合文本航空信息和轨迹数据，提升航班延误预测精度，展示出文本与轨迹数据互补的优势。

2606.07528 2026-06-09 cs.CL cs.AI cs.LG 新提交 91%

BEACON: Behavioral Entropy Aggregation for Cross-Model Hallucination Detection in Large Language Models

BEACON: 面向大语言模型跨模型幻觉检测的行为熵聚合

Naveen Bera, Pulijala Sai Nikhila, Kondaguduru Abhiram, Shaik Gayaz Ali, Shoaib Sadiq Salehmohamed, Shaik Mohammed Omar, Jinal Prashant Thakkar, Hansika Aredla, Shalmali Ayachit

机构 * LLM Lens

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,abstract_cn)；分类 cs.CL、cs.AI、cs.LG

AI总结提出BEACON框架，通过多维度行为特征（语义熵、嵌入几何、思维链一致性、释义稳定性）的黑盒检测方法，在7个基准上达到0.8123 AUROC，优于现有方法。

Comments 12 pages, 6 tables, 1 figure. Code and data available upon request

2501.00885 2026-05-04 cs.CL cs.AI cs.LG 91%

Representation in large language models

大型语言模型中的表示

Cameron Yetman

机构 * University of Toronto（多伦多大学）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,abstract_cn)；分类 cs.CL、cs.AI、cs.LG

AI总结本文探讨大型语言模型是否通过基于表示的信息处理而非记忆和随机查找来工作，提出通过实践技术研究表示并发展解释，为未来理论提供基础。

Comments Preprint, forthcoming in Ergo: An Open Access Journal of Philosophy, 34 pages, 2 figures

2604.15109 2026-04-21 cs.CL cs.AI cs.LG 91%

IUQ: Interrogative Uncertainty Quantification for Long-Form Large Language Model Generation

IUQ：长文本大语言模型生成的疑问性不确定性量化

Haozhi Fan, Jinhao Duan, Kaidi Xu

机构 * University of Pennsylvania（宾夕法尼亚大学）； UNC Chapel Hill（北卡罗来纳大学教堂山分校）； City University of Hong Kong（香港城市大学）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,abstract_cn)；分类 cs.CL、cs.AI、cs.LG

AI总结本文提出IUQ框架，通过跨样本一致性与内样本忠实度量化长文本生成的不确定性，验证了其在不同模型上的优越性能。

Comments Accepted by ACL (main) 2026

2604.13991 2026-04-16 cs.CL cs.AI cs.LG 91%

Adaptive Conformal Prediction for Improving Factuality of Generations by Large Language Models

自适应置信区间预测用于通过大语言模型提高事实性

Aleksandr Rubashevskii, Dzianis Piatrashyn, Preslav Nakov, Maxim Panov

机构 * Mohamed bin Zayed University of Artificial Intelligence（莫扎德·本·扎耶德人工智能大学）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,abstract_cn)；分类 cs.CL、cs.AI、cs.LG

AI总结本文提出自适应置信区间预测方法，用于提升大语言模型生成内容的事实性，通过改进置信度校准以提高条件覆盖性能。

2602.04297 2026-02-05 cs.CL cs.AI cs.LG 91%

Revisiting Prompt Sensitivity in Large Language Models for Text Classification: The Role of Prompt Underspecification

重新审视大型语言模型在文本分类中的提示敏感性：提示不足的作用

Branislav Pecher, Michal Spiegel, Robert Belanec, Jan Cegin

机构 * Kempelen Institute of Intelligent Technologies（智能技术研究所）； Faculty of Information Technology, Brno University of Technology（信息科技学院）； Faculty of Informatics, Masaryk University（信息学院）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；prompting(abstract)

AI总结本文研究了大型语言模型在文本分类中提示敏感性的来源，发现提示不足是导致性能波动的主要原因，并通过实验分析揭示了其对模型内部表示的影响。

2510.12032 2025-10-15 cs.CL cs.AI cs.LG 91%

Multi-stage Prompt Refinement for Mitigating Hallucinations in Large Language Models

Jung-Woo Shim, Yeong-Joon Ju, Ji-Hoon Park, Seong-Whan Lee

机构 * Department of Artificial Intelligence, Korea University（人工智能系，韩国大学）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；small language model(abstract)

Comments 22 pages, 6 figures

2504.17004 2025-06-03 cs.LG cs.AI cs.CL stat.ML 91%

(Im)possibility of Automated Hallucination Detection in Large Language Models

Amin Karbasi, Omar Montasser, John Sous, Grigoris Velegkas

机构 * Yale University（耶鲁大学）

专题命中知识编辑与模型理解：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；RLHF(abstract)

1. 知识编辑与模型理解 7505 篇

Understanding and Mitigating Prompt Leaking Attacks in Real-World LLM-Based Applications

Abstention-Aware Personalized Object Rearrangement via Uncertainty-Guided LLM Assistance

Belief at Risk: Quantifying Agentic AI Model Risk with LLM-Inferred Bayesian State Filters

Phoneme-First Prediction for LLM-Based Speech Recognition

Decentralized LLM-Driven Coordination of Acoustic Robots for Contactless Object Manipulation

UVMarvel: an Automated LLM-aided UVM Machine for Subsystem-level RTL Verification

From Natural Language to Silicon: The Representation Bottleneck in LLM Hardware Design

Persona Alchemy: Designing, Evaluating, and Implementing Psychologically-Grounded LLM Agents for Diverse Stakeholder Representation

Rethinking Visual Prompting for Multimodal Large Language Models with External Knowledge

Large Language Models Know Your Contextual Search Intent: A Prompting Framework for Conversational Search

Behavior and Representation in Open-Weight Large Language Models for Combinatorial Optimization: From Feature Extraction to Algorithm Selection

AURORA: Asymmetry and Update-Induced Rotation for Robust Hallucination Detection in Large Language Models

CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models

Text-Attributed Knowledge Graph Enrichment with Large Language Models for Medical Concept Representation

Psychological Steering of Large Language Models

Quantifying Uncertainty in Natural Language Explanations of Large Language Models

Beyond Hooking Onto the World: Referential Profiles and the Numerical Structure of LLM Grounding

Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography

Prompting Large Language Models for Training-Free Non-Intrusive Load Monitoring

Enhancing Multi-Agent Consensus through Third-Party LLM Integration: Analyzing Uncertainty and Mitigating Hallucinations in Large Language Models

Improving Large Language Model (LLM) fidelity through context-aware grounding: A systematic approach to reliability and veracity

Small Language Model enabled Autonomous agent for Language-Conditioned Cognitive Radar

LLM4Delay: Flight Delay Prediction via Cross-Modality Adaptation of Large Language Models and Aircraft Trajectory Representation

BEACON: Behavioral Entropy Aggregation for Cross-Model Hallucination Detection in Large Language Models

Representation in large language models

IUQ: Interrogative Uncertainty Quantification for Long-Form Large Language Model Generation

Adaptive Conformal Prediction for Improving Factuality of Generations by Large Language Models

Revisiting Prompt Sensitivity in Large Language Models for Text Classification: The Role of Prompt Underspecification

Multi-stage Prompt Refinement for Mitigating Hallucinations in Large Language Models

(Im)possibility of Automated Hallucination Detection in Large Language Models