自适应方法论文趋势 2023–2026

主要结论

• 自适应方法论文从 2023 年的 143 篇增长到 2026 年的 410 篇，三年间扩大到约 2.9 倍。
• NeurIPS 是发自适应方法论文最多的会议，2023–2026 年共 365 篇，占四个会议总量的 36%。
• 交叉方向里涨得最快的是理论 · optimization：同时属于两个主题的论文从 2023 年的 1 篇增至 2026 年的 29 篇。
• 在「优化与缩放」的 5 个主题里，它的 2026 年论文数排第 3；同组增长最快的缩放定律是 4.1 倍，自适应方法是 2.9 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 2026AcceleGrad#: Adaptive Geometry-Aware Acceleration
• ICML 2026Adaptive Bandit Algorithms for Contextual Matching Markets
• ICLR 2026Adaptive Width Neural Networks
• CVPR 2026Active Inference for Micro-Gesture Recognition EFE-Guided Temporal Sampling and Adaptive
• NeurIPS 2025AC-DiT: Adaptive Coordination Diffusion Transformer for Mobile Manipulation
• ICML 2025Adaptive Data Collection for Robust Learning Across Multiple Distributions
• ICLR 2025Adaptive Length Image Tokenization via Recurrent Allocation
• CVPR 2025A Universal Scale-Adaptive Deformable Transformer for Image Restoration across Diverse Artifacts
• NeurIPS 2024Semi-Discrete Optimal Transport: Nearly Minimax Estimation With Stochastic Gradient Descent and Adaptive Entropic Regularization
• ICML 2024A New Linear Scaling Rule for Private Adaptive Hyperparameter Optimization
• ICLR 2024Learning Adaptive Multiresolution Transforms via Meta-Framelet-based Graph Convolutional Network
• CVPR 2024HybridNeRF: Efficient Neural Rendering via Adaptive Volumetric Surfaces
• NeurIPS 2023A Bounded Ability Estimation for Computerized Adaptive Testing
• ICML 2023A/B Testing in Network Data with Covariate-Adaptive Randomization
• ICLR 2023An Adaptive Policy to Employ Sharpness-Aware Minimization
• CVPR 2023Adaptive Annealing for Robust Geometric Estimation
• NeurIPS 2026AdapMatch: Adaptive Bias Decoupling for Semi-Supervised Partial Label Learning under Unknown Class Distributions
• ICML 2026Adaptive Batch Sizes Using Non-Euclidean Gradient Noise Scales for Stochastic Sign and Spectral Descent
• ICLR 2026ARMOR: High-Performance Semi-Structured Pruning via Adaptive Matrix Factorization
• CVPR 2026Adaptive Learned Image Compression with Graph Neural Networks

全部自适应方法论文

共 24 篇，第 1 / 1 页

1. ICLR 2023Adaptive Budget Allocation for Parameter-Efficient Fine-Tuning
2. ICLR 2023Adaptive Optimization in the $\infty$-Width Limit
3. ICLR 2023An Adaptive Policy to Employ Sharpness-Aware Minimization
4. ICLR 2023Boosting Causal Discovery via Adaptive Sample Reweighting
5. ICLR 2023Continual Pre-training of Language Models
6. ICLR 2023Deep Learning meets Nonparametric Regression: Are Weight-Decayed DNNs Locally Adaptive?
7. ICLR 2023Differentially Private Adaptive Optimization with Delayed Preconditioners
8. ICLR 2023Diffusion-GAN: Training GANs with Diffusion
9. ICLR 2023Distributed Extra-gradient with Optimal Complexity and Communication Guarantees
10. ICLR 2023DySR: Adaptive Super-Resolution via Algorithm and System Co-design
11. ICLR 2023Fast and Precise: Adjusting Planning Horizon with Adaptive Subgoal Search
12. ICLR 2023FreeMatch: Self-adaptive Thresholding for Semi-supervised Learning
13. ICLR 2023Imbalanced Semi-supervised Learning with Bias Adaptive Classifier
14. ICLR 2023Is Adversarial Training Really a Silver Bullet for Mitigating Data Poisoning?
15. ICLR 2023MaskFusion: Feature Augmentation for Click-Through Rate Prediction via Input-adaptive Mask Fusion
16. ICLR 2023Maximizing Communication Efficiency for Large-scale Training via 0/1 Adam
17. ICLR 2023Neural ePDOs: Spatially Adaptive Equivariant Partial Differential Operator Based Networks
18. ICLR 2023Noise Is Not the Main Factor Behind the Gap Between Sgd and Adam on Transformers, But Sign Descent Might Be
19. ICLR 2023Pruning Deep Neural Networks from a Sparsity Perspective
20. ICLR 2023Robust Algorithms on Adaptive Inputs from Bounded Adversaries
21. ICLR 2023SP2 : A Second Order Stochastic Polyak Method
22. ICLR 2023TiAda: A Time-scale Adaptive Algorithm for Nonconvex Minimax Optimization
23. ICLR 2023When to Make and Break Commitments?
24. ICLR 2023Win: Weight-Decay-Integrated Nesterov Acceleration for Adaptive Gradient Algorithms

常见问题

2023–2026 年四大会议一共有多少篇自适应方法论文？

共 1,012 篇，按年份分别是 2023 年 143 篇、2024 年 183 篇、2025 年 276 篇、2026 年 410 篇。

NeurIPS 2026 有多少篇自适应方法论文？

NeurIPS 2026 有 138 篇自适应方法论文，上一届是 109 篇。

ICLR 2026 有多少篇自适应方法论文？

ICLR 2026 有 80 篇自适应方法论文，上一届是 47 篇。

ICML 2026 有多少篇自适应方法论文？

ICML 2026 有 109 篇自适应方法论文，上一届是 68 篇。

CVPR 2026 有多少篇自适应方法论文？

CVPR 2026 有 83 篇自适应方法论文，上一届是 52 篇。

哪个会议的自适应方法论文最多？

NeurIPS，2023–2026 年共 365 篇。