大模型推理能力 - arXivDaily 专题

2311.17365 2023-11-30 cs.CV 78%

Symbol-LLM: Leverage Language Models for Symbolic System in Visual Human Activity Reasoning

Xiaoqian Wu, Yong-Lu Li, Jianhua Sun, Cewu Lu

专题命中逻辑推理：reasoning(title,abstract)

Comments Accepted by NeurIPS 2023

2311.04901 2023-11-09 cs.CV 78%

GENOME: GenerativE Neuro-symbOlic visual reasoning by growing and reusing ModulEs

Zhenfang Chen, Rui Sun, Wenjun Liu, Yining Hong, Chuang Gan

专题命中逻辑推理：reasoning(title,abstract)

2303.01571 2023-03-06 cs.CC 78%

Complexity of Reasoning with Cardinality Minimality Conditions

Nadia Creignou, Frédéric Olive, Johannes Schmidt

专题命中逻辑推理：reasoning(title,abstract)

2303.01379 2023-03-03 eess.SY cs.RO cs.SY 78%

Planning and Control of Uncertain Cooperative Mobile Manipulator-Endowed Systems under Temporal-Logic Tasks

Christos Verginis

专题命中逻辑推理：planning(title,abstract)

Comments PhD thesis; also available at http://kth.diva-portal.org/smash/get/diva2:1427745/FULLTEXT01.pdf

2102.04361 2021-03-10 cs.FL cs.LO cs.MA 78%

Regular Model Checking Approach to Knowledge Reasoning over Parameterized Systems (technical report)

Daniel Stan, Anthony Widjaja Lin

专题命中逻辑推理：reasoning(title,abstract)

Comments Extended version, version of record accepted at the 20th International Conference on Autonomous Agents and Multiagent Systems (AAMAS-21)

2012.07912 2020-12-16 cs.RO 78%

Reactive Temporal Logic Planning for Multiple Robots in Unknown Occupancy Grid Maps

Yiannis Kantaros, Matthew Malencia, George J. Pappas

专题命中逻辑推理：planning(title,abstract)

2011.00992 2020-11-03 stat.OT 78%

The P-T Probability Framework for Semantic Communication, Falsification, Confirmation, and Bayesian Reasoning

Chenguang Lu

专题命中逻辑推理：reasoning(title,abstract)

Comments 36 pages; 10 Figures

Journal ref Philosophies 2020, 5(4), 25

1911.05908 2019-11-15 cs.MA cs.LO 78%

Tractable reasoning about Agent Programming in Dynamic Preference Logic

Marlo Souza, Álvaro Moreira, Renata Vieira

专题命中逻辑推理：reasoning(title,abstract)

Comments Published in BRACIS 2018

1911.05907 2019-11-15 cs.MA cs.LO 78%

A Dynamic Preference Logic for reasoning about Agent Programming

Marlo Souza, Álvaro Moreira, Renata Vieira, John-Jules Ch. Meyer

专题命中逻辑推理：reasoning(title,abstract)

Comments piblished on BRACIS 2017

1907.09101 2019-07-23 cs.LO cs.MA 78%

When Do Introspection Axioms Matter for Multi-Agent Epistemic Reasoning?

Yifeng Ding, Wesley H. Holliday, Cedegao Zhang

专题命中逻辑推理：reasoning(title,abstract)

Comments In Proceedings TARK 2019, arXiv:1907.08335

Journal ref EPTCS 297, 2019, pp. 121-139

1907.09100 2019-07-23 cs.GT 78%

Reasoning about Social Choice and Games in Monadic Fixed-Point Logic

Ramit Das, R. Ramanujam, Sunil Simon

专题命中逻辑推理：reasoning(title,abstract)

Comments In Proceedings TARK 2019, arXiv:1907.08335

Journal ref EPTCS 297, 2019, pp. 106-120

1906.09836 2019-06-25 cs.RO 78%

Learning Grasp Affordance Reasoning through Semantic Relations

Paola Ardón, Èric Pairet, Ronald P. A. Petrick, Subramanian Ramamoorthy, Katrin S. Lohan

专题命中逻辑推理：reasoning(title,abstract)

Comments Accepted in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2019

1905.11537 2019-05-29 cs.LO 78%

Reasoning about Quality and Fuzziness of Strategic Behaviours

Patricia Bouyer, Orna Kupferman, Nicolas Markey, Bastien Maubert, Aniello Murano, Giuseppe Perelli

专题命中逻辑推理：reasoning(title,abstract)

1905.06627 2019-05-17 cs.LO 78%

Reasoning about Cognitive Trust in Stochastic Multiagent Systems

Xiaowei Huang, Marta Kwiatkowska, Maciej Olejnik

专题命中逻辑推理：reasoning(title,abstract)

Comments to be published in TOCL ACM journal

1611.08541 2019-03-14 cs.LO 78%

Reasoning about Strategies: on the Satisfiability Problem

Fabio Mogavero, Aniello Murano, Giuseppe Perelli, Moshe Y. Vardi

专题命中逻辑推理：reasoning(title,abstract)

Comments arXiv admin note: text overlap with arXiv:1112.6275, arXiv:1202.1309

Journal ref Logical Methods in Computer Science, Volume 13, Issue 1 (March 17, 2017) lmcs:3204

1812.11741 2019-01-01 cs.LO 78%

A modal aleatoric calculus for probabilistic reasoning: extended version

Tim French, Andrew Gozzard, Mark Reynolds

专题命中逻辑推理：reasoning(title,abstract)

Comments Long version of paper accepted to appear at the 2019 Indian Conference on Logic and Applictaions

1408.1647 2014-08-08 cs.LO 78%

Computing consensus: A logic for reasoning about deliberative processes based on argumentation

Truls Pedersen, Sjur Dyrkolbotn

专题命中逻辑推理：reasoning(title,abstract)

Comments Presented at the 1st International Workshop on Argument for Agreement and Assurance (AAA 2013)

1112.6275 2014-02-13 cs.LO cs.MA math.LO 78%

Reasoning About Strategies: On the Model-Checking Problem

Fabio Mogavero, Aniello Murano, Giuseppe Perelli, Moshe Y. Vardi

专题命中逻辑推理：reasoning(title,abstract)

1006.0220 2010-06-02 cs.LO cs.CC 78%

The Complexity of Reasoning for Fragments of Autoepistemic Logic

Nadia Creignou, Arne Meier, Michael Thomas, Heribert Vollmer

专题命中逻辑推理：reasoning(title,abstract)

0903.2448 2009-12-01 cs.LO cs.MA 78%

Positive Logic with Adjoint Modalities: Proof Theory, Semantics and Reasoning about Information

Mehrnoosh Sadrzadeh, Roy Dyckhoff

专题命中逻辑推理：reasoning(title,abstract)

Comments This paper is the full version of the article that is to appear in the ENTCS proceedings of the 25th conference on the Mathematical Foundations of Programming Semantics (MFPS), April 2009, University of Oxford

cs/0603020 2009-12-01 cs.LO cs.MA 78%

Reasoning About Knowledge of Unawareness

Joseph Y. halpern, Leandro Chaves Rego

专题命中逻辑推理：reasoning(title,abstract)

Comments 32 pages

2608.12486 2026-08-14 cs.CL 新提交 77%

DIVE: Unlocking Self-Improvement in Frozen Language Models Through Diversity-Driven Skill Evolution

DIVE：通过多样性驱动的技能进化解锁冻结语言模型的自我提升

Siheng Xiong, Ali Payani, Oguzhan Gungordu, Faramarz Fekri

专题命中逻辑推理：reasoning(abstract)；logical reasoning(abstract)；verifier(abstract)；分类 cs.CL

AI总结 DIVE是一种多样性驱动的无参数框架，可让冻结LLM从任务经验和验证器反馈中进化出自然语言技能，在多项推理任务上优于现有方法，还能实现模型间技能迁移，让小模型性能匹配或超越大模型。

2601.03808 2026-08-06 cs.CV cs.LG 版本更新 77%

From Brute Force to Semantic Insight: Performance-Guided Data Transformation Design with LLMs

从暴力到语义洞察：基于LLM的性能引导数据转换设计

Usha Shrestha, Dmitry Ignatov, Radu Timofte

机构 * Computer Vision Lab, CAIDAS & IFI, University of Würzburg（计算机视觉实验室、CAIDAS与IFI、乌尔姆大学）

专题命中逻辑推理：chain-of-thought(abstract,abstract_cn)；reasoning(abstract)；分类 cs.LG

AI总结本文提出了一种基于LLM的性能引导数据转换设计方法，通过经验反馈实现闭环优化，减少穷举搜索需求，提升代码生成的准确性与任务对齐性。

2607.12982 2026-07-22 cs.AI cs.MA cs.SC 版本更新 77%

FormalAnalyticGeo: A Neural-Symbolic Based Framework for Multimodal Analytic Geometry Problem Generation

形式分析几何：一种基于神经符号的多模态解析几何问题生成框架

Ruoran Xu, Wending Gao, Qiufeng Wang

机构 * Xi’an Jiaotong-Liverpool University（西交利物浦大学）

专题命中逻辑推理：reasoning(abstract)；math reasoning(abstract)；verifier(abstract)；分类 cs.AI

AI总结研究解析几何问题生成，提出基于神经符号的FormalAnalyticGeo框架，利用CDL及四个大语言模型组件，无需人工注释自动生成问题，形成闭环，生成的AnalyticGeo7K数据集问题误差小，框架和数据集将公开。

2602.14095 2026-07-07 cs.AI cs.CR 版本更新 77%

NEST: Nascent Encoded Steganographic Thoughts

NEST：新生编码隐写思想

Artem Karpov

专题命中逻辑推理：reasoning(abstract)；chain-of-thought(abstract)；CoT(abstract)；分类 cs.AI

AI总结探讨模型在无害文本中隐藏秘密推理的隐写思维链，通过分类系统评估34个模型的隐写能力极限，测量相关指标并与基线比较，发现前沿模型无法联合推理嵌入，编码能力已达标，强调持续评估隐写风险的必要性。

2604.20811 2026-04-23 cs.AI 77%

Diagnosing CFG Interpretation in LLMs

在LLM中诊断CFG解释

Hanqi Li, Lu Chen, Kai Yu

机构 * X-LANCE Lab, School of Computer Science, Shanghai Jiao Tong University, Shanghai, China（上海交通大学计算机科学学院X-LANCE实验室）； AISpeech Co., Ltd., Suzhou, China（上海AI语音有限公司）； Shanghai Innovation Institution, Shanghai, China（上海创新研究所）； Jiangsu Key Lab of Language Computing, Suzhou, China（江苏省语言计算重点实验室）； Suzhou Laboratory, Suzhou, China（苏州实验室）

专题命中逻辑推理：CoT(abstract,abstract_cn)；reasoning(abstract)；分类 cs.AI

AI总结研究LLM在处理新上下文无关文法时能否生成语法正确、行为功能和语义忠实的输出，揭示LLM在语法、行为和语义层面的层次退化问题。

2512.02304 2026-04-22 cs.CL 77%

When Does Verification Pay Off? A Closer Look at LLMs as Solution Verifiers

验证何时有效？对LLM作为解决方案验证器的深入探讨

Jack Lu, Ryan Teehan, Jinran Jin, Mengye Ren

机构 * Agentic Learning AI Lab, New York University（代理学习人工智能实验室，纽约大学）

专题命中逻辑推理：reasoning(abstract)；logical reasoning(abstract)；verifier(abstract)；分类 cs.CL

AI总结研究探讨了在何种条件下验证有效，通过分析37个模型在9个基准测试中的表现，发现跨模型家族验证效果优于自验证，且验证收益随模型相似性增加而降低。

Comments Accepted at ICLR 2026 AI with Recursive Self-Improvement workshop

2604.17225 2026-04-21 cs.CL 77%

A Multi-Agent Approach for Claim Verification from Tabular Data Documents

从表格数据文档中验证声明的多智能体方法

Rudra Ranajee Saha, Laks V. S. Lakshmanan, Raymond T. Ng

机构 * University of British Columbia（不列颠哥伦比亚大学）

专题命中逻辑推理：reasoning(abstract)；chain-of-thought(abstract)；verifier(abstract)；分类 cs.CL

AI总结本文提出多智能体框架MACE，通过规划器、执行器和验证器三个智能体，实现可解释的表格数据验证，实验显示其在多个数据集上达到SOTA性能，且在参数较少时仍表现优异。

2604.04869 2026-04-07 cs.LG 77%

Optimizing LLM Prompt Engineering with DSPy Based Declarative Learning

基于DSPy的声明式学习优化大语言模型提示工程

Shiek Ruksana, Sailesh Kiran Kurra, Thipparthi Sanjay Baradwaj

机构 * Vasavi College of Engineering（瓦萨维工程学院）； Amazon（亚马逊）； Texas（德克萨斯州）

专题命中逻辑推理：reasoning(abstract)；chain-of-thought(abstract)；planning(abstract)；分类 cs.LG

AI总结本文提出基于DSPy的声明式学习方法，通过符号规划、无梯度优化和自动模块重写提升提示优化的可靠性、效率和泛化能力，实验显示事实准确率提升30-45%，幻觉率降低25%。

Comments Best paper Award ,IEEE International Conference on Emerging Smart Computing and Informatics (ESCI) Pune, India. Mar 11-13, 2026

2603.25498 2026-03-27 cs.AI 77%

EcoThink: A Green Adaptive Inference Framework for Sustainable and Accessible Agents

EcoThink: 一种绿色自适应推理框架，用于可持续和可及的智能体

Linxiao Li, Zhixiang Lu

机构 * The University of Sydney（悉尼大学）； University of Liverpool（利物浦大学）

专题命中逻辑推理：reasoning(abstract)；chain-of-thought(abstract)；CoT(abstract)；分类 cs.AI

AI总结本文提出EcoThink框架，通过轻量级路由器动态评估查询复杂度，减少推理能耗40.4%，提升AI可持续性和可及性。

Comments Accepted by WWW 2026

1. 逻辑推理 3016 篇

Symbol-LLM: Leverage Language Models for Symbolic System in Visual Human Activity Reasoning

GENOME: GenerativE Neuro-symbOlic visual reasoning by growing and reusing ModulEs

Complexity of Reasoning with Cardinality Minimality Conditions

Planning and Control of Uncertain Cooperative Mobile Manipulator-Endowed Systems under Temporal-Logic Tasks

Regular Model Checking Approach to Knowledge Reasoning over Parameterized Systems (technical report)

Reactive Temporal Logic Planning for Multiple Robots in Unknown Occupancy Grid Maps

The P-T Probability Framework for Semantic Communication, Falsification, Confirmation, and Bayesian Reasoning

Tractable reasoning about Agent Programming in Dynamic Preference Logic

A Dynamic Preference Logic for reasoning about Agent Programming

When Do Introspection Axioms Matter for Multi-Agent Epistemic Reasoning?

Reasoning about Social Choice and Games in Monadic Fixed-Point Logic

Learning Grasp Affordance Reasoning through Semantic Relations

Reasoning about Quality and Fuzziness of Strategic Behaviours

Reasoning about Cognitive Trust in Stochastic Multiagent Systems

Reasoning about Strategies: on the Satisfiability Problem

A modal aleatoric calculus for probabilistic reasoning: extended version

Computing consensus: A logic for reasoning about deliberative processes based on argumentation

Reasoning About Strategies: On the Model-Checking Problem

The Complexity of Reasoning for Fragments of Autoepistemic Logic

Positive Logic with Adjoint Modalities: Proof Theory, Semantics and Reasoning about Information

Reasoning About Knowledge of Unawareness

DIVE: Unlocking Self-Improvement in Frozen Language Models Through Diversity-Driven Skill Evolution

From Brute Force to Semantic Insight: Performance-Guided Data Transformation Design with LLMs

FormalAnalyticGeo: A Neural-Symbolic Based Framework for Multimodal Analytic Geometry Problem Generation

NEST: Nascent Encoded Steganographic Thoughts

Diagnosing CFG Interpretation in LLMs

When Does Verification Pay Off? A Closer Look at LLMs as Solution Verifiers

A Multi-Agent Approach for Claim Verification from Tabular Data Documents

Optimizing LLM Prompt Engineering with DSPy Based Declarative Learning

EcoThink: A Green Adaptive Inference Framework for Sustainable and Accessible Agents