数学推理论文趋势 2023–2026

主要结论

• 数学推理论文从 2023 年的 57 篇增长到 2026 年的 577 篇，三年间扩大到约 10 倍。
• NeurIPS 是发数学推理论文最多的会议，2023–2026 年共 469 篇，占四个会议总量的 43%。
• 交叉方向里涨得最快的是强化学习：同时属于两个主题的论文从 2023 年的 3 篇增至 2026 年的 140 篇。
• 在「神经符号」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的可微推理是 2.5 倍，数学推理是 10 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026A mathematical theory of balancing relational generalization and memorization
• ICML 2026ASyMOB: Algebraic Symbolic Mathematical Operations Benchmark
• ICLR 2026AutoGPS: Automated Geometry Problem Solving via Multimodal Formalization and Deductive Reasoning
• CVPR 2026Affine Perspective-Three-Point Problem
• NeurIPS 2025Are Language Models Efficient Reasoners? A Perspective from Logic Programming
• ICML 2025Autoformulation of Mathematical Optimization Models Using LLMs
• ICLR 2025Advancing Mathematical Reasoning in Language Models: The Impact of Problem-Solving Data, Data Synthesis Methods, and Training Stages
• CVPR 2025Bridging Viewpoint Gaps: Geometric Reasoning Boosts Semantic Correspondence
• NeurIPS 2024A Careful Examination of Large Language Model Performance on Grade School Arithmetic
• ICML 2024Accelerating Legacy Numerical Solvers by Non-intrusive Gradient-based Meta-solving
• ICLR 2024Don't Trust: Verify -- Grounding LLM Quantitative Reasoning with Autoformalization
• CVPR 2024E-GPS: Explainable Geometry Problem Solving via Top-Down Solver and Bottom-Up Generator
• NeurIPS 2023Evaluating and Improving Tool-Augmented Computation-Intensive Math Reasoning
• ICML 2023Automated Search for Conjectures on Mathematical Constants using Analysis of Integer Sequences
• ICLR 2023Differentiable Mathematical Programming for Object-Centric Representation Learning
• CVPR 2023Abstract Visual Reasoning: An Algebraic Approach for Solving Raven's Progressive Matrices
• NeurIPS 2026A Mechanistic Analysis of Looped Reasoning Language Models
• ICML 2026Beyond Theorem Proving: Formulation, Framework and Benchmark for Formal Problem-Solving
• ICLR 2026CogFlow: Bridging Perception and Reasoning through Knowledge Internalization for Visual Mathematical Problem Solving
• CVPR 2026Hilbert-Geo Solving Solid Geometric Problems by Neural-Symbolic Reasoning

全部数学推理论文

共 314 篇，第 3 / 7 页

101. ICML 2025MA-LoT: Model-Collaboration Lean-based Long Chain-of-Thought Reasoning enhances Formal Theorem Proving
102. ICML 2025Machine Learning meets Algebraic Combinatorics: A Suite of Datasets Capturing Research-level Conjecturing Ability in Pure Mathematics
103. ICML 2025Machines and Mathematical Mutations: Using GNNs to Characterize Quiver Mutation Classes
104. ICML 2025MARGE: Improving Math Reasoning with Guided Exploration
105. ICML 2025MASS: Mathematical Data Selection via Skill Graphs for Pretraining Large Language Models
106. ICML 2025MATH-Perturb: Benchmarking LLMs' Math Reasoning Abilities against Hard Perturbations
107. ICML 2025MathConstruct: Challenging LLM Reasoning with Constructive Proofs
108. ICML 2025Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation
109. ICML 2025MME-CoT: Benchmarking Chain-of-Thought in Large Multimodal Models for Reasoning Quality, Robustness, and Efficiency
110. ICML 2025One Example Shown, Many Concepts Known! Counterexample-Driven Conceptual Reasoning in Mathematical LLMs
111. ICML 2025Position: AI's growing due process problem
112. ICML 2025Position: Formal Mathematical Reasoning—A New Frontier in AI
113. ICML 2025Premise-Augmented Reasoning Chains Improve Error Identification in Math reasoning with LLMs
114. ICML 2025Putnam-AXIOM: A Functional & Static Benchmark for Measuring Higher Level Mathematical Reasoning in LLMs
115. ICML 2025Regress, Don't Guess: A Regression-like Loss on Number Tokens for Language Models
116. ICML 2025rStar-Math: Small LLMs Can Master Math Reasoning with Self-Evolved Deep Thinking
117. ICML 2025Solving Satisfiability Modulo Counting Exactly with Probabilistic Circuits
118. ICML 2025STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving
119. ICML 2025T1: Advancing Language Model Reasoning through Reinforcement Learning and Inference Scaling
120. ICML 2025Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning
121. ICML 2025unMORE: Unsupervised Multi-Object Segmentation via Center-Boundary Reasoning
122. ICML 2025VersaPRM: Multi-Domain Process Reward Model via Synthetic Reasoning Data
123. ICML 2025What Do Learning Dynamics Reveal About Generalization in LLM Mathematical Reasoning?
124. ICML 2025What Makes In-context Learning Effective for Mathematical Reasoning
125. NeurIPS 20253D-Prover: Diversity Driven Theorem Proving With Determinantal Point Processes
126. NeurIPS 2025A learnability analysis on neuro-symbolic learning
127. NeurIPS 2025A Theoretical Study on Bridging Internal Probability and Self-Consistency for LLM Reasoning
128. NeurIPS 2025A*-Thought: Efficient Reasoning via Bidirectional Compression for Low-Resource Settings
129. NeurIPS 2025Absolute Zero: Reinforced Self-play Reasoning with Zero Data
130. NeurIPS 2025AbstentionBench: Reasoning LLMs Fail on Unanswerable Questions
131. NeurIPS 2025Accelerating RL for LLM Reasoning with Optimal Advantage Regression
132. NeurIPS 2025AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning
133. NeurIPS 2025Act Only When It Pays: Efficient Reinforcement Learning for LLM Reasoning via Selective Rollouts
134. NeurIPS 2025Activation-Guided Consensus Merging for Large Language Models
135. NeurIPS 2025Ada-R1: Hybrid-CoT via Bi-Level Adaptive Reasoning Optimization
136. NeurIPS 2025AdaReasoner: Adaptive Reasoning Enables More Flexible Thinking
137. NeurIPS 2025Agentic RL Scaling Law: Spontaneous Code Execution for Mathematical Problem Solving
138. NeurIPS 2025Anchored Diffusion Language Model
139. NeurIPS 2025AOR: Anatomical Ontology-Guided Reasoning for Medical Large Multimodal Model in Chest X-Ray Interpretation
140. NeurIPS 2025APOLLO: Automated LLM and Lean Collaboration for Advanced Formal Reasoning
141. NeurIPS 2025Are Language Models Efficient Reasoners? A Perspective from Logic Programming
142. NeurIPS 2025AREAL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoning
143. NeurIPS 2025ARM: Adaptive Reasoning Model
144. NeurIPS 2025AtmosSci-Bench: Evaluating the Recent Advance of Large Language Model for Atmospheric Science
145. NeurIPS 2025Atomic Thinking of LLMs: Decoupling and Exploring Mathematical Reasoning Abilities
146. NeurIPS 2025Auditing Meta-Cognitive Hallucinations in Reasoning Large Language Models
147. NeurIPS 2025AutoJudge: Judge Decoding Without Manual Annotation
148. NeurIPS 2025AutoOpt: A Dataset and a Unified Framework for Automating Optimization Problem Solving
149. NeurIPS 2025Bag of Tricks for Inference-time Computation of LLM Reasoning
150. NeurIPS 2025Benchmarking Large Language Models with Integer Sequence Generation Tasks

常见问题

2023–2026 年四大会议一共有多少篇数学推理论文？

共 1,085 篇，按年份分别是 2023 年 57 篇、2024 年 137 篇、2025 年 314 篇、2026 年 577 篇。

NeurIPS 2026 有多少篇数学推理论文？

NeurIPS 2026 有 208 篇数学推理论文，上一届是 190 篇。

ICLR 2026 有多少篇数学推理论文？

ICLR 2026 有 175 篇数学推理论文，上一届是 61 篇。

ICML 2026 有多少篇数学推理论文？

ICML 2026 有 113 篇数学推理论文，上一届是 45 篇。

CVPR 2026 有多少篇数学推理论文？

CVPR 2026 有 81 篇数学推理论文，上一届是 18 篇。

哪个会议的数学推理论文最多？

NeurIPS，2023–2026 年共 469 篇。