变分推断论文趋势 2023–2026

主要结论

• 变分推断论文从 2023 年的 106 篇降到 2026 年的 91 篇，热度在回落。
• NeurIPS 是发变分推断论文最多的会议，2023–2026 年共 124 篇，占四个会议总量的 35%。
• 交叉方向里涨得最快的是采样方法：同时属于两个主题的论文从 2023 年的 1 篇增至 2026 年的 6 篇。
• 在「概率与贝叶斯」的 5 个主题里，它的 2026 年论文数排第 5；同组增长最快的高斯过程是 2.2 倍，变分推断是 0.9 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 Reproducible Evaluation Protocol for Quantum Non-Linearity in Variational Quantum Models
• ICML 2026Beyond Independence: Learning Correlated Views for Variational Incomplete Multi-View Clustering
• ICLR 2026Optimizing Data Augmentation through Bayesian Model Selection
• CVPR 2026BrepVGAE Variational Graph Autoencoder with Unified Latent Representation for B-rep
• NeurIPS 2025Covariate-moderated Empirical Bayes Matrix Factorization
• ICML 2025A Variational Information Theoretic Approach to Out-of-Distribution Detection
• ICLR 2025Bridging the Gap between Variational Inference and Stochastic Gradient MCMC in Function Space
• CVPR 2025Dora: Sampling and Benchmarking for 3D Shape Variational Auto-Encoders
• NeurIPS 2024Training Binary Neural Networks via Gaussian Variational Inference and Low-Rank Semidefinite Programming
• ICML 2024Amortized Variational Deep Kernel Learning
• ICLR 2024Causal Inference with Conditional Front-Door Adjustment and Identifiable Variational Autoencoder
• CVPR 2024WorDepth: Variational Language Prior for Monocular Depth Estimation
• NeurIPS 2023[Re] Variational Neural Cellular Automata
• ICML 2023Coupled Variational Autoencoder
• ICLR 2023A VAE for Transformers with Nonparametric Variational Information Bottleneck
• CVPR 2023Confidence-Aware Personalized Federated Learning via Variational Expectation Maximization
• NeurIPS 2026Adaptive Delayed-Update Cyclic Algorithm for Variational Inequalities
• ICML 2026Dual-stage Contrastive Learning-enhanced Multi-view Variational Clustering
• ICLR 2026Pareto Variational Autoencoder
• CVPR 2026Variational Graph-based Normal Integration

全部变分推断论文

共 359 篇，第 3 / 8 页

101. ICLR 2025Injective flows for star-like manifolds
102. ICLR 2025Microcanonical Langevin Ensembles: Advancing the Sampling of Bayesian Neural Networks
103. ICLR 2025Nonlinear multiregion neural dynamics with parametric impulse response communication channels
104. ICLR 2025Nonlinear Sequence Embedding by Monotone Variational Inequality
105. ICLR 2025On the Relation between Trainability and Dequantization of Variational Quantum Learning Models
106. ICLR 2025Random-Set Neural Networks
107. ICLR 2025Repulsive Latent Score Distillation for Solving Inverse Problems
108. ICLR 2025Residual Deep Gaussian Processes on Manifolds
109. ICLR 2025Risk-Sensitive Variational Actor-Critic: A Model-Based Approach
110. ICLR 2025Scalable Bayesian Learning with posteriors
111. ICLR 2025SoftCVI: Contrastive variational inference with self-generated soft labels
112. ICLR 2025Solving hidden monotone variational inequalities with surrogate losses
113. ICLR 2025Stochastic variance-reduced Gaussian variational inference on the Bures-Wasserstein manifold
114. ICLR 2025Support is All You Need for Certified VAE Training
115. ICLR 2025Token Statistics Transformer: Linear-Time Attention via Variational Rate Reduction
116. ICLR 2025VAE-Var: Variational Autoencoder-Enhanced Variational Methods for Data Assimilation in Meteorology
117. ICLR 2025Variational Bayesian Pseudo-Coreset
118. ICLR 2025Variational Best-of-N Alignment
119. ICLR 2025Variational Diffusion Posterior Sampling with Midpoint Guidance
120. ICLR 2025Variational Search Distributions
121. ICLR 2025When do GFlowNets learn the right distribution?
122. ICML 2025$K^2$VAE: A Koopman-Kalman Enhanced Variational AutoEncoder for Probabilistic Time Series Forecasting
123. ICML 2025A Variational Information Theoretic Approach to Out-of-Distribution Detection
124. ICML 2025Aggregation of Dependent Expert Distributions in Multimodal Variational Autoencoders
125. ICML 2025Be a Goldfish: Forgetting Bad Conditioning in Sparse Linear Regression via Variational Autoencoders
126. ICML 2025Censor Dependent Variational Inference
127. ICML 2025Continuous Semi-Implicit Models
128. ICML 2025Cross-Modal Alignment via Variational Copula Modelling
129. ICML 2025Deep Bayesian Filter for Bayes-Faithful Data Assimilation
130. ICML 2025Expected Variational Inequalities
131. ICML 2025Federated Generalised Variational Inference: A Robust Probabilistic Federated Learning Framework
132. ICML 2025Flexible Tails for Normalizing Flows
133. ICML 2025Global-Local Dirichlet Processes for Clustering Grouped Data in the Presence of Group-Specific Idiosyncratic Variables
134. ICML 2025Identifying Causal Direction via Variational Bayesian Compression
135. ICML 2025Layer-wise Quantization for Quantized Optimistic Dual Averaging
136. ICML 2025Learning Likelihood-Free Reference Priors
137. ICML 2025Multilayer Matrix Factorization via Dimension-Reducing Diffusion Variational Inference
138. ICML 2025Neural Guided Diffusion Bridges
139. ICML 2025New Bounds for Sparse Variational Gaussian Processes
140. ICML 2025Pareto-frontier Entropy Search with Variational Lower Bound Maximization
141. ICML 2025Revisiting Unbiased Implicit Variational Inference
142. ICML 2025Score Matching with Missing Data
143. ICML 2025Synonymous Variational Inference for Perceptual Image Compression
144. ICML 2025Tensor-Var: Efficient Four-Dimensional Variational Data Assimilation
145. ICML 2025Variational Counterfactual Intervention Planning to Achieve Target Outcomes
146. ICML 2025Variational Learning of Fractional Posteriors
147. ICML 2025Variational Phylogenetic Inference with Products over Bipartitions
148. ICML 2025Variational Rectified Flow Matching
149. ICML 2025VCT: Training Consistency Models with Variational Noise Coupling
150. NeurIPS 2025Amortized Variational Transdimensional Inference

常见问题

2023–2026 年四大会议一共有多少篇变分推断论文？

共 359 篇，按年份分别是 2023 年 106 篇、2024 年 76 篇、2025 年 86 篇、2026 年 91 篇。

NeurIPS 2026 有多少篇变分推断论文？

NeurIPS 2026 有 30 篇变分推断论文，上一届是 28 篇。

ICLR 2026 有多少篇变分推断论文？

ICLR 2026 有 23 篇变分推断论文，上一届是 28 篇。

ICML 2026 有多少篇变分推断论文？

ICML 2026 有 35 篇变分推断论文，上一届是 28 篇。

CVPR 2026 有多少篇变分推断论文？

CVPR 2026 有 3 篇变分推断论文，上一届是 2 篇。

哪个会议的变分推断论文最多？

NeurIPS，2023–2026 年共 124 篇。