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 113 篇，第 1 / 3 页

1. ICML 2026Advancing LLM Reasoning with Natural Language and Numerical Feedback
2. ICML 2026Agentic Proposing: Enhancing Large language Model Reasoning via Compositional Skill Synthesis
3. ICML 2026Anchored Policy Optimization: Mitigating Exploration Collapse via Support-Constrained Rectification
4. ICML 2026Are Large Reasoning Models Interruptible?
5. ICML 2026ASyMOB: Algebraic Symbolic Mathematical Operations Benchmark
6. ICML 2026Batched Contextual Reinforcement
7. ICML 2026Batched First-Order Methods for Parallel LP Solving in MIP
8. ICML 2026BeaconKV: Key-Value Cache Compression Guided by Beacon Queries for Efficient Large Reasoning Model Inference
9. ICML 2026Beyond Mode Collapse: Distribution Matching for Diverse Reasoning
10. ICML 2026Beyond the Final Answer: Evaluating the Reasoning Trajectories of Tool-Augmented Agents
11. ICML 2026Beyond Theorem Proving: Formulation, Framework and Benchmark for Formal Problem-Solving
12. ICML 2026Bridging On-Device and Cloud LLMs for Collaborative Reasoning: A Unified Methodology for Local Routing and Post-Training
13. ICML 2026Characterizing, Evaluating, and Optimizing Complex Reasoning
14. ICML 2026Conformal Path Reasoning: Trustworthy Knowledge Graph Question Answering via Path-Level Calibration
15. ICML 2026CoRe: Collaborative Reasoning via Cross Teaching
16. ICML 2026Correcting in Hindsight: Editing Past Key-Value States for Robust LLM Reasoning
17. ICML 2026CPMöbius: Iterative Coach–Player Reasoning for Data-Free Reinforcement Learning
18. ICML 2026Critique-Guided Distillation for Robust Reasoning via Refinement
19. ICML 2026CRPO: Character-centric Group Relative Policy Optimization for Role-aware Reasoning in Role-playing Agents
20. ICML 2026D-ARL: A Distribution-Matched Asynchronous Reinforcement Learning Framework for Language Reasoning
21. ICML 2026Demystifying Scientific Problem-Solving in LLMs by Probing Knowledge and Reasoning
22. ICML 2026DenseSteer: Steering Small Language Models towards Dense Math Reasoning
23. ICML 2026Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution
24. ICML 2026Differential Smoothing Mitigates Sharpening and Improves LLM Reasoning
25. ICML 2026Does Math Reasoning Improve General LLM Capabilities? Understanding Transferability of LLM Reasoning
26. ICML 2026Don't Ignore the Tail: Decoupled Distillation Produces Top Maths Students on an Academic Budget
27. ICML 2026DyCon: Dynamic Reasoning Control via Evolving Difficulty Modeling
28. ICML 2026Dynamic Thinking-Token Selection for Efficient Reasoning in Large Reasoning Models
29. ICML 2026Dynamics Reveals Structure: Challenging the Linear Propagation Assumption
30. ICML 2026Editable Proof Sketch for Automated Theorem Proving
31. ICML 2026Effective Reasoning Chains Reduce Intrinsic Dimensionality
32. ICML 2026Emergent Analogical Reasoning in Transformers
33. ICML 2026Equilibrium Reasoners: Learning Attractors Enables Scalable Reasoning
34. ICML 2026Evaluating AI Grading on Real-World Handwritten College Mathematics: A Large-Scale Study Toward a Benchmark
35. ICML 2026Evaluating LLMs When They Do Not Know the Answer: Statistical Evaluation of Mathematical Reasoning via Comparative Signals
36. ICML 2026Evaluating Relational Reasoning in LLMs with REL
37. ICML 2026Evidential Reasoning Advances Interpretable Real-World Disease Screening
38. ICML 2026Experience Augmented Policy Optimization for LLM Reasoning
39. ICML 2026Faults in Our Formal Benchmarking: Dataset Defects and Evaluation Failures in Lean Theorem Proving
40. ICML 2026Feature Resemblance: Towards a Theoretical Understanding of Analogical Reasoning in Transformers
41. ICML 2026FloorplanQA: A Benchmark for Spatial Reasoning in LLMs using Structured Representations
42. ICML 2026FormalRx: Rectify and eXamine Semantic Failures in Autoformalization
43. ICML 2026From LLM-Generated Conjectures to Lean Formalizations: Automated Polynomial Inequality Proving via Sum-of-Squares Certificates
44. ICML 2026From Seeing to Thinking: Decoupling Perception and Reasoning Improves Post-Training of Vision-Language Models
45. ICML 2026From Shortcuts to Reasoning: Robust Post-Training of Theory of Mind with Reinforcement Learning
46. ICML 2026Geometry of Reason: Spectral Signatures of Valid Mathematical Reasoning
47. ICML 2026Hallucination Detection from Structural Reasoning Model
48. ICML 2026HERMES: Towards Efficient and Verifiable Mathematical Reasoning in LLMs
49. ICML 2026Hierarchical Reinforcement Learning for Sparse-Reward Search in Commutative Algebra
50. ICML 2026How Far Ahead Do LLMs Plan? Uncovering the Latent Horizon in Chain-of-Thought Reasoning

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。