因果发现论文趋势 2023–2026

主要结论

• 因果发现论文从 2023 年的 59 篇增长到 2026 年的 139 篇，三年间扩大到约 2.4 倍。
• NeurIPS 是发因果发现论文最多的会议，2023–2026 年共 150 篇，占四个会议总量的 40%。
• 交叉方向里涨得最快的是通用机器学习 · unsupervised and semisupervised learning：同时属于两个主题的论文从 2023 年的 3 篇增至 2026 年的 10 篇。
• 在「因果与公平」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的偏见缓解是 1.5 倍，因果发现是 2.4 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 Characterization of Latent Variable Causal Models Consistent with Observational Data
• ICML 2026A Recursive Decomposition Framework for Causal Structure Learning in the Presence of Latent Variables
• ICLR 2026Bures Generalized Category Discovery
• CVPR 2026Learning by Analogy A Causal Framework for Compositional Generalization
• NeurIPS 2025Causal Differentiating Concepts: Interpreting LM Behavior via Causal Representation Learning
• ICML 2025Bayesian Active Learning for Bivariate Causal Discovery
• ICLR 2025A Conditional Independence Test in the Presence of Discretization
• CVPR 2025Hyperbolic Category Discovery
• NeurIPS 2024Causal Bandits for Linear Structural Equation Models
• ICML 2024Adaptive Online Experimental Design for Causal Discovery
• ICLR 2024A Versatile Causal Discovery Framework to Allow Causally-Related Hidden Variables
• CVPR 2024Contrastive Mean-Shift Learning for Generalized Category Discovery
• NeurIPS 2023A Scale-Invariant Sorting Criterion to Find a Causal Order in Additive Noise Models
• ICML 2023Active causal structure learning with advice
• ICLR 2023Boosting Causal Discovery via Adaptive Sample Reweighting
• CVPR 2023Intrinsic Physical Concepts Discovery With Object-Centric Predictive Models
• NeurIPS 2026Act, Validate, Adapt: Closing the Causal Discovery-Control Loop
• ICML 2026Causal Discovery for Irregularly Time Series with Consistency Guarantees
• ICLR 2026Causal Discovery in the Wild: A Voting-Theoretic Ensemble Approach
• CVPR 2026PlanaReLoc Camera Relocalization in 3D Planar Primitives via Region-Based Structure

全部因果发现论文

共 379 篇，第 2 / 8 页

51. ICML 2026Evaluating Bivariate Causal Statements Based on Mutual Compatibility
52. ICML 2026Formalizing and Falsifying Causal Pathways of Rare Events
53. ICML 2026Function-Valued Causal Influence in Nonlinear Time Series
54. ICML 2026Global Directional Priors with Local Statistical Validation for Scalable Causal Discovery
55. ICML 2026Hierarchical Causal Abduction: A Foundation Framework for Explainable Model Predictive Control
56. ICML 2026Identifiable Markov Switching Models with Instantaneous Effects and Exponential Families
57. ICML 2026Identifiable Nonlinear Differentiable Causal Discovery via Independence and Adaptive Group Sparsity
58. ICML 2026Identifying Partially Observed Causal Models from Heterogeneous/Nonstationary Data
59. ICML 2026Interpretable Neural ODEs for Gene Regulatory Network Discovery under Perturbations
60. ICML 2026Learning General Causal Structures with Hidden Dynamic Process for Climate Analysis
61. ICML 2026Linear Causal Representation Learning by Topological Ordering, Pruning, and Disentanglement
62. ICML 2026Local Covariate Selection for Average Causal Effect Estimation without Pretreatment and Causal Sufficiency Assumptions
63. ICML 2026Mind Dreamer: Untethering Imagination via Active Causal Intervention on Latent Manifolds
64. ICML 2026Multi-View Causal Discovery without Non-Gaussianity: Identifiability and Algorithms
65. ICML 2026On the Identifiability of Poisson Branching Structural Causal Model Under Latent Confounding
66. ICML 2026One Intervention per Component is Enough: Towards Identifiability in Linear Stochastic Dynamics from Steady State
67. ICML 2026PACER: Acyclic Causal Discovery from Large-scale Interventional Data
68. ICML 2026PathwayLLM: Explainable Clinical Trajectory Modeling with Structured Pathways for Sepsis Prediction
69. ICML 2026Power-Boosted Granger-Causal Discovery for Large Heterogeneous Panel Data
70. ICML 2026Powerful and Theoretically Guaranteed Independence Testing on Heterogeneous Federated Clients
71. ICML 2026Regret-Based Federated Causal Discovery with Unknown Interventions
72. ICML 2026Ripple Perturbations Through Structure: Likelihood-Constrained Adversarial Attacks on Heterogeneous Tabular Data
73. ICML 2026Robust Causal Discovery in Real-World Time Series with Power-Laws
74. ICML 2026Root Cause Analysis of Failures in Microservices via Bayesian Root Cause Discovery
75. ICML 2026SCOUT: Cyclic Causal Discovery Under Soft Interventions with Unknown Targets
76. ICML 2026Shapley Regularized Neural Granger Causality
77. ICML 2026Skewness-Robust Causal Discovery in Location-Scale Noise Models
78. ICML 2026Spatial Deconfounder: Interference-Aware Deconfounding for Spatial Causal Inference
79. ICML 2026Spiked-CFR: Causal Representation Learning from LLMs via Wasserstein Projection Pursuit
80. ICML 2026Split Group Knockoffs: Controlling False Discovery Rate in Transformational Group Sparsity
81. ICML 2026Test-Time Learning of Causal Structure from Interventional Data
82. ICML 2026The Fisher Dimension: Instance-Dependent Complexity for Causal Discovery
83. ICML 2026TopoDistill: Distilling Global System Topology for Causal Discovery in Multivariate Time Series
84. ICML 2026Towards Completeness in Causal Discovery from Soft Interventions with Known Targets
85. ICML 2026Unveiling the Structure of Do-Calculus Reasoning via Derivation Graphs
86. ICML 2026Use What You Know: Causal Foundation Models with Partial Graphs
87. NeurIPS 2026A Characterization of Latent Variable Causal Models Consistent with Observational Data
88. NeurIPS 2026Act, Validate, Adapt: Closing the Causal Discovery-Control Loop
89. NeurIPS 2026AgentCBO: Causal Belief Routing for Bayesian Optimization under Unknown Graphs
90. NeurIPS 2026Attention-Based Pretraining for Unsupervised Amortized Causal Discovery
91. NeurIPS 2026Automated Causal Effect Estimation through Self-Evolving AI
92. NeurIPS 2026Causal Abstractions, Categorically Unified
93. NeurIPS 2026Causal Discovery from Unseen Environments
94. NeurIPS 2026Causal discovery needs explicit epistemic standards
95. NeurIPS 2026Causal Discovery Under Hard Selection Bias: A New Robust Score-Matching Approach
96. NeurIPS 2026Causal Discovery under Time-Varying Delays
97. NeurIPS 2026Causal Discovery via Transformed Low-Rank Quantile Surfaces
98. NeurIPS 2026Causal Discovery with False Positive Error Control
99. NeurIPS 2026Causal Gradient Steering: Exposing Shortcut Gradients by Destroying Causal Signal
100. NeurIPS 2026Causal learning with the invariance principle

常见问题

2023–2026 年四大会议一共有多少篇因果发现论文？

共 379 篇，按年份分别是 2023 年 59 篇、2024 年 82 篇、2025 年 99 篇、2026 年 139 篇。

NeurIPS 2026 有多少篇因果发现论文？

NeurIPS 2026 有 53 篇因果发现论文，上一届是 34 篇。

ICLR 2026 有多少篇因果发现论文？

ICLR 2026 有 25 篇因果发现论文，上一届是 27 篇。

ICML 2026 有多少篇因果发现论文？

ICML 2026 有 54 篇因果发现论文，上一届是 31 篇。

CVPR 2026 有多少篇因果发现论文？

CVPR 2026 有 7 篇因果发现论文，上一届是 7 篇。

哪个会议的因果发现论文最多？

NeurIPS，2023–2026 年共 150 篇。