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 1,085 篇，第 18 / 22 页

851. NeurIPS 2025Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Space
852. NeurIPS 2025SolidGeo: Measuring Multimodal Spatial Math Reasoning in Solid Geometry
853. NeurIPS 2025SolverLLM: Leveraging Test-Time Scaling for Optimization Problem via LLM-Guided Search
854. NeurIPS 2025Solving Inequality Proofs with Large Language Models
855. NeurIPS 2025Solving the Asymmetric Traveling Salesman Problem via Trace-Guided Cost Augmentation
856. NeurIPS 2025SPC: Evolving Self-Play Critic via Adversarial Games for LLM Reasoning
857. NeurIPS 2025SPRINT: Enabling Interleaved Planning and Parallelized Execution in Reasoning Models
858. NeurIPS 2025SQL-R1: Training Natural Language to SQL Reasoning Model By Reinforcement Learning
859. NeurIPS 2025STNet: Spectral Transformation Network for Solving Operator Eigenvalue Problem
860. NeurIPS 2025Stop Summation: Min-Form Credit Assignment Is All Process Reward Model Needs for Reasoning
861. NeurIPS 2025Structured Linear CDEs: Maximally Expressive and Parallel-in-Time Sequence Models
862. NeurIPS 2025SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
863. NeurIPS 2025SwS: Self-aware Weakness-driven Problem Synthesis in Reinforcement Learning for LLM Reasoning
864. NeurIPS 2025SynLogic: Synthesizing Verifiable Reasoning Data at Scale for Learning Logical Reasoning and Beyond
865. NeurIPS 2025System-1.5 Reasoning: Traversal in Language and Latent Spaces with Dynamic Shortcuts
866. NeurIPS 2025TAPAS: Datasets for Learning the Learning with Errors Problem
867. NeurIPS 2025Teaching Language Models to Reason with Tools
868. NeurIPS 2025The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity
869. NeurIPS 2025The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning
870. NeurIPS 2025The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning
871. NeurIPS 2025Think Only When You Need with Large Hybrid-Reasoning Models
872. NeurIPS 2025Think or Not? Exploring Thinking Efficiency in Large Reasoning Models via an Information-Theoretic Lens
873. NeurIPS 2025Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains
874. NeurIPS 2025Thinker: Learning to Think Fast and Slow
875. NeurIPS 2025Thoughts Are All Over the Place: On the Underthinking of Long Reasoning Models
876. NeurIPS 2025TokenSqueeze: Performance-Preserving Compression for Reasoning LLMs
877. NeurIPS 2025Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Properties
878. NeurIPS 2025Towards Thinking-Optimal Scaling of Test-Time Compute for LLM Reasoning
879. NeurIPS 2025Training Language Models to Reason Efficiently
880. NeurIPS 2025Trajectory Bellman Residual Minimization: A Simple Value-Based Method for LLM Reasoning
881. NeurIPS 2025Transformers Provably Learn Chain-of-Thought Reasoning with Length Generalization
882. NeurIPS 2025Tropical Attention: Neural Algorithmic Reasoning for Combinatorial Algorithms
883. NeurIPS 2025Trust, But Verify: A Self-Verification Approach to Reinforcement Learning with Verifiable Rewards
884. NeurIPS 2025Uni-LoRA: One Vector is All You Need
885. NeurIPS 2025Value-Guided Search for Efficient Chain-of-Thought Reasoning
886. NeurIPS 2025VeriThinker: Learning to Verify Makes Reasoning Model Efficient
887. NeurIPS 2025WHAT MAKES MATH PROBLEMS HARD FOR REINFORCEMENT LEARNING: A CASE STUDY
888. NeurIPS 2025When No Paths Lead to Rome: Benchmarking Systematic Neural Relational Reasoning
889. NeurIPS 2025When Thinking Fails: The Pitfalls of Reasoning for Instruction-Following in LLMs
890. NeurIPS 2025Which Data Attributes Stimulate Math and Code Reasoning? An Investigation via Influence Functions
891. NeurIPS 2025Who Reasons in the Large Language Models?
892. CVPR 2024A&B BNN: Add&Bit-Operation-Only Hardware-Friendly Binary Neural Network
893. CVPR 2024Beyond First-Order Tweedie: Solving Inverse Problems using Latent Diffusion
894. CVPR 2024E-GPS: Explainable Geometry Problem Solving via Top-Down Solver and Bottom-Up Generator
895. CVPR 2024Generating Handwritten Mathematical Expressions From Symbol Graphs: An End-to-End Pipeline
896. CVPR 2024Higher-order Relational Reasoning for Pedestrian Trajectory Prediction
897. CVPR 2024Know Your Neighbors: Improving Single-View Reconstruction via Spatial Vision-Language Reasoning
898. CVPR 2024MoReVQA: Exploring Modular Reasoning Models for Video Question Answering
899. CVPR 2024PixelLM: Pixel Reasoning with Large Multimodal Model
900. CVPR 2024Solving the Catastrophic Forgetting Problem in Generalized Category Discovery

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。