反事实解释论文趋势 2023–2026

主要结论

• 反事实解释论文从 2023 年的 21 篇增长到 2026 年的 26 篇，三年间扩大到约 1.2 倍。
• NeurIPS 是发反事实解释论文最多的会议，2023–2026 年共 31 篇，占四个会议总量的 34%。
• 交叉方向里涨得最快的是反事实公平：同时属于两个主题的论文从 2023 年的 8 篇增至 2026 年的 10 篇。
• 在「可解释性」的 5 个主题里，它的 2026 年论文数排第 4；同组增长最快的机制解释是 8.1 倍，反事实解释是 1.2 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 2026Offloading Score: Measuring AI Reliance through Counterfactual Workflows
• ICML 2026Density-Guided Robust Counterfactual Explanations on Tabular Data under Model Multiplicity
• ICLR 2026Training-free Counterfactual Explanation for Temporal Graph Model Inference
• CVPR 2026Back to the Feature Explaining Video Classifiers with Video Counterfactual
• NeurIPS 2025Causal Explanation-Guided Learning for Organ Allocation
• ICML 2025Exogenous Isomorphism for Counterfactual Identifiability
• ICLR 2025Generating Likely Counterfactuals Using Sum-Product Networks
• CVPR 2025Towards Fine-Grained Interpretability: Counterfactual Explanations for Misclassification with Saliency Partition
• NeurIPS 2024AR-Pro: Counterfactual Explanations for Anomaly Repair with Formal Properties
• ICML 2024CF-OPT: Counterfactual Explanations for Structured Prediction
• ICLR 2024Counterfactual Density Estimation using Kernel Stein Discrepancies
• CVPR 2024Looking Similar Sounding Different: Leveraging Counterfactual Cross-Modal Pairs for Audiovisual Representation Learning
• NeurIPS 2023[Re] FOCUS: Flexible Optimizable Counterfactual Explanations for Tree Ensembles
• ICML 2023Counterfactual Analysis in Dynamic Latent State Models
• ICLR 2023Measuring Forgetting of Memorized Training Examples
• CVPR 2023OCTET: Object-Aware Counterfactual Explanations
• NeurIPS 2026Tracing Actual Causes with Counterfactual Witness Maps
• ICML 2026Algorithmic Recourse of In-Context Learning for Tabular Data
• ICLR 2026ATEX-CF: Attack-Informed Counterfactual Explanations for Graph Neural Networks
• CVPR 2026Making the Classification Explanation Faithful to the Confidence Score

全部反事实解释论文

共 21 篇，第 1 / 1 页

1. CVPR 2023Adversarial Counterfactual Visual Explanations
2. CVPR 2023Grounding Counterfactual Explanation of Image Classifiers to Textual Concept Space
3. CVPR 2023IDGI: A Framework To Eliminate Explanation Noise From Integrated Gradients
4. CVPR 2023Masked Images Are Counterfactual Samples for Robust Fine-Tuning
5. CVPR 2023OCTET: Object-Aware Counterfactual Explanations
6. CVPR 2023Spatial-Temporal Concept Based Explanation of 3D ConvNets
7. CVPR 2023Towards Transferable Targeted Adversarial Examples
8. ICLR 2023Measuring Forgetting of Memorized Training Examples
9. ICLR 2023Robust Explanation Constraints for Neural Networks
10. ICML 2023Cluster Explanation via Polyhedral Descriptions
11. ICML 2023Counterfactual Analysis in Dynamic Latent State Models
12. ICML 2023Eventual Discounting Temporal Logic Counterfactual Experience Replay
13. ICML 2023GLOBE-CE: A Translation Based Approach for Global Counterfactual Explanations
14. ICML 2023PWSHAP: A Path-Wise Explanation Model for Targeted Variables
15. ICML 2023Robust Counterfactual Explanations for Neural Networks With Probabilistic Guarantees
16. ICML 2023Sequential Counterfactual Risk Minimization
17. ICML 2023Trustworthy Policy Learning under the Counterfactual No-Harm Criterion
18. NeurIPS 2023[Re] FOCUS: Flexible Optimizable Counterfactual Explanations for Tree Ensembles
19. NeurIPS 2023Counterfactual Evaluation of Peer-Review Assignment Policies
20. NeurIPS 2023Learning with Explanation Constraints
21. NeurIPS 2023The Utility of “Even if” Semifactual Explanation to Optimise Positive Outcomes

常见问题

2023–2026 年四大会议一共有多少篇反事实解释论文？

共 90 篇，按年份分别是 2023 年 21 篇、2024 年 20 篇、2025 年 23 篇、2026 年 26 篇。

NeurIPS 2026 有多少篇反事实解释论文？

NeurIPS 2026 有 15 篇反事实解释论文，上一届是 7 篇。

ICLR 2026 有多少篇反事实解释论文？

ICLR 2026 有 5 篇反事实解释论文，上一届是 5 篇。

ICML 2026 有多少篇反事实解释论文？

ICML 2026 有 4 篇反事实解释论文，上一届是 9 篇。

CVPR 2026 有多少篇反事实解释论文？

CVPR 2026 有 2 篇反事实解释论文，上一届是 2 篇。

哪个会议的反事实解释论文最多？

NeurIPS，2023–2026 年共 31 篇。