深度学习 · optimization论文趋势 2023–2026

主要结论

• 深度学习 · optimization论文从 2023 年的 256 篇降到 2026 年的 240 篇，热度在回落。
• NeurIPS 是发深度学习 · optimization论文最多的会议，2023–2026 年共 500 篇，占四个会议总量的 52%。
• 交叉方向里涨得最快的是对齐与安全：同时属于两个主题的论文从 2023 年的 5 篇增至 2026 年的 23 篇。
• 在「深度学习」的 11 个主题里，它的 2026 年论文数排第 6；同组增长最快的扩散模型是 4.1 倍，深度学习 · optimization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 2026Achieving Directional-Stationarity from a Single Random Direction Step
• ICML 2026Bottleneck Communication Delay Minimization for Communication-Efficient Decentralized Learning
• ICLR 2026From Sorting Algorithms to Scalable Kernels: Bayesian Optimization in High-Dimensional Permutation Spaces
• CVPR 2026Coverage Optimization for Camera View Selection
• NeurIPS 2025High-Performance Arithmetic Circuit Optimization via Differentiable Architecture Search
• ICML 2025Feature-Mapping Topology Optimization with Neural Heaviside Signed Distance Functions
• ICLR 2025Boosting Neural Combinatorial Optimization for Large-Scale Vehicle Routing Problems
• CVPR 2025HotSpot: Signed Distance Function Optimization with an Asymptotically Sufficient Condition
• NeurIPS 2024$\boldsymbol{\mu}\mathbf{P^2}$: Effective Sharpness Aware Minimization Requires Layerwise Perturbation Scaling
• ICML 2024Barrier Algorithms for Constrained Non-Convex Optimization
• ICLR 2024Bayesian Coreset Optimization for Personalized Federated Learning
• CVPR 2024Multi-Session SLAM with Differentiable Wide-Baseline Pose Optimization
• NeurIPS 2023$\textbf{A}^2\textbf{CiD}^2$: Accelerating Asynchronous Communication in Decentralized Deep Learning
• ICML 2023Buying Information for Stochastic Optimization
• ICLR 2023Alternating Differentiation for Optimization Layers
• CVPR 2023Visual-Language Prompt Tuning With Knowledge-Guided Context Optimization
• NeurIPS 2026Adaptive Conditional Gradient Sliding: Projection-Free and Line-Search-Free Acceleration
• ICML 2026Riemannian Optimization for Fair Spectral Clustering
• ICLR 2026Gradient-Based Diversity Optimization with Differentiable Top-$k$ Objective
• CVPR 2026Its Never Too Late Noise Optimization for Collapse Recovery in

全部深度学习 · optimization论文

共 72 篇，第 2 / 2 页

51. ICML 2024Optimization without Retraction on the Random Generalized Stiefel Manifold
52. ICML 2024Plug-in Performative Optimization
53. ICML 2024Position: Leverage Foundational Models for Black-Box Optimization
54. ICML 2024Position: Optimization in SciML Should Employ the Function Space Geometry
55. ICML 2024Posterior Sampling-Based Bayesian Optimization with Tighter Bayesian Regret Bounds
56. ICML 2024Principled Preferential Bayesian Optimization
57. ICML 2024Probability Distribution of Hypervolume Improvement in Bi-objective Bayesian Optimization
58. ICML 2024Projection-Free Online Convex Optimization with Time-Varying Constraints
59. ICML 2024Quantum Algorithm for Online Exp-concave Optimization
60. ICML 2024Robust Data-driven Prescriptiveness Optimization
61. ICML 2024Robust Optimization in Protein Fitness Landscapes Using Reinforcement Learning in Latent Space
62. ICML 2024Sample Average Approximation for Conditional Stochastic Optimization with Dependent Data
63. ICML 2024Skill Set Optimization: Reinforcing Language Model Behavior via Transferable Skills
64. ICML 2024Stochastic Optimization with Arbitrary Recurrent Data Sampling
65. ICML 2024Stochastic Weakly Convex Optimization beyond Lipschitz Continuity
66. ICML 2024Symmetric Replay Training: Enhancing Sample Efficiency in Deep Reinforcement Learning for Combinatorial Optimization
67. ICML 2024Token-level Direct Preference Optimization
68. ICML 2024Tuning-Free Stochastic Optimization
69. ICML 2024Two-timescale Derivative Free Optimization for Performative Prediction with Markovian Data
70. ICML 2024Universal Gradient Methods for Stochastic Convex Optimization
71. ICML 2024Vanilla Bayesian Optimization Performs Great in High Dimensions
72. ICML 2024Zeroth-Order Methods for Constrained Nonconvex Nonsmooth Stochastic Optimization

常见问题

2023–2026 年四大会议一共有多少篇深度学习 · optimization论文？

共 953 篇，按年份分别是 2023 年 256 篇、2024 年 294 篇、2025 年 163 篇、2026 年 240 篇。

NeurIPS 2026 有多少篇深度学习 · optimization论文？

NeurIPS 2026 有 103 篇深度学习 · optimization论文，上一届是 62 篇。

ICLR 2026 有多少篇深度学习 · optimization论文？

ICLR 2026 有 44 篇深度学习 · optimization论文，上一届是 51 篇。

ICML 2026 有多少篇深度学习 · optimization论文？

ICML 2026 有 86 篇深度学习 · optimization论文，上一届是 44 篇。

CVPR 2026 有多少篇深度学习 · optimization论文？

CVPR 2026 有 7 篇深度学习 · optimization论文，上一届是 6 篇。

哪个会议的深度学习 · optimization论文最多？

NeurIPS，2023–2026 年共 500 篇。