验证论文趋势 2023–2026

主要结论

• 验证论文从 2023 年的 8 篇增长到 2026 年的 48 篇，三年间扩大到约 6.0 倍。
• NeurIPS 是发验证论文最多的会议，2023–2026 年共 45 篇，占四个会议总量的 41%。
• 交叉方向里涨得最快的是推理模型：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 8 篇。
• 在「安全与可信」的 5 个主题里，它的 2026 年论文数排第 4；同组增长最快的水印是 9.3 倍，验证是 6.0 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 2026Affine-Image Propagation for Tight and Scalable $\ell_{2}$ Neural Network Verification
• ICML 2026Euclean: Automated Geometry Problem Formalization with Unified Verification in Lean
• ICLR 2026Parameterized Hardness of Zonotope Containment and Neural Network Verification
• CVPR 2026Goldilocks Test Sets for Face Verification
• NeurIPS 2025Self-Verification Provably Prevents Model Collapse in Recursive Synthetic Training
• ICML 2025SDP-CROWN: Efficient Bound Propagation for Neural Network Verification with Tightness of Semidefinite Programming
• ICLR 2025Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification
• CVPR 2025PHGC: Procedural Heterogeneous Graph Completion for Natural Language Task Verification in Egocentric Videos
• NeurIPS 2024Critically Assessing the State of the Art in Neural Network Verification
• ICML 2024Compact Optimality Verification for Optimization Proxies
• ICLR 2024CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing
• CVPR 2024Towards General Robustness Verification of MaxPool-based Convolutional Neural Networks via Tightening Linear Approximation
• NeurIPS 2023[Re] On the Reproducibility of “FairCal: Fairness Calibration for Face Verification”
• ICML 2023Trapdoor Normalization with Irreversible Ownership Verification
• ICLR 2023Fundamental Limits in Formal Verification of Message-Passing Neural Networks
• CVPR 2023Efficient Verification of Neural Networks Against LVM-Based Specifications
• NeurIPS 2026End-to-End Verification of Neuro-symbolic Automata via Contrastive Logit-Gaps
• ICML 2026The Latent Guardian: Defending Collaborative Perception via Feature-Level Consistency Verification
• ICLR 2026Test-time Verification via Optimal Transport: Coverage, ROC, & Sub-optimality
• CVPR 2026Lipschitz Optimization for Formal Verification of Homographies

全部验证论文

共 33 篇，第 1 / 1 页

1. CVPR 2025Neuro-Symbolic Evaluation of Text-to-Video Models using Formal Verification
2. CVPR 2025PHGC: Procedural Heterogeneous Graph Completion for Natural Language Task Verification in Egocentric Videos
3. CVPR 2025Tightening Robustness Verification of MaxPool-based Neural Networks via Minimizing the Over-Approximation Zone
4. CVPR 2025Training Data Provenance Verification: Did Your Model Use Synthetic Data from My Generative Model for Training?
5. ICLR 2025Beyond Model Collapse: Scaling Up with Synthesized Data Requires Verification
6. ICLR 2025Block Verification Accelerates Speculative Decoding
7. ICLR 2025Data Taggants: Dataset Ownership Verification Via Harmless Targeted Data Poisoning
8. ICLR 2025Rethinking and Improving Autoformalization: Towards a Faithful Metric and a Dependency Retrieval-based Approach
9. ICLR 2025Sensitivity Verification for Additive Decision Tree Ensembles
10. ICLR 2025Triples as the Key: Structuring Makes Decomposition and Verification Easier in LLM-based TableQA
11. ICML 2025AlphaVerus: Bootstrapping Formally Verified Code Generation through Self-Improving Translation and Treefinement
12. ICML 2025DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts
13. ICML 2025No Soundness in the Real World: On the Challenges of the Verification of Deployed Neural Networks
14. ICML 2025PatchPilot: A Cost-Efficient Software Engineering Agent with Early Attempts on Formal Verification
15. ICML 2025Position: Trustworthy AI Agents Require the Integration of Large Language Models and Formal Methods
16. ICML 2025SDP-CROWN: Efficient Bound Propagation for Neural Network Verification with Tightness of Semidefinite Programming
17. ICML 2025Solving Probabilistic Verification Problems of Neural Networks using Branch and Bound
18. ICML 2025Verification Learning: Make Unsupervised Neuro-Symbolic System Feasible
19. NeurIPS 2025Characterizing the Expressivity of Fixed-Precision Transformer Language Models
20. NeurIPS 2025Clip-and-Verify: Linear Constraint-Driven Domain Clipping for Accelerating Neural Network Verification
21. NeurIPS 2025Compositional Neural Network Verification via Assume-Guarantee Reasoning
22. NeurIPS 2025DSCS: Fast CPDAG-Based Verification of Collapsible Submodels in High-Dimensional Bayesian Networks
23. NeurIPS 2025Efficiently Verifiable Proofs of Data Attribution
24. NeurIPS 2025Generating and Checking DNN Verification Proofs
25. NeurIPS 2025Learning Simple Interpolants for Linear Integer Arithmetic
26. NeurIPS 2025Let a Neural Network be Your Invariant
27. NeurIPS 2025Limitations of Normalization in Attention
28. NeurIPS 2025RvLLM: LLM Runtime Verification with Domain Knowledge
29. NeurIPS 2025Self-Verification Provably Prevents Model Collapse in Recursive Synthetic Training
30. NeurIPS 2025SpecEM: Training-Free LLM Ensembling via Iterative Drafting, Verification, and Online Feedback
31. NeurIPS 2025Traversal Verification for Speculative Tree Decoding
32. NeurIPS 2025Two‑Stage Learning of Stabilizing Neural Controllers via Zubov Sampling and Iterative Domain Expansion
33. NeurIPS 2025Weaver: Shrinking the Generation-Verification Gap by Scaling Compute for Verification

常见问题

2023–2026 年四大会议一共有多少篇验证论文？

共 111 篇，按年份分别是 2023 年 8 篇、2024 年 22 篇、2025 年 33 篇、2026 年 48 篇。

NeurIPS 2026 有多少篇验证论文？

NeurIPS 2026 有 19 篇验证论文，上一届是 15 篇。

ICLR 2026 有多少篇验证论文？

ICLR 2026 有 14 篇验证论文，上一届是 6 篇。

ICML 2026 有多少篇验证论文？

ICML 2026 有 12 篇验证论文，上一届是 8 篇。

CVPR 2026 有多少篇验证论文？

CVPR 2026 有 3 篇验证论文，上一届是 4 篇。

哪个会议的验证论文最多？

NeurIPS，2023–2026 年共 45 篇。