Theory · learning theory papers, 2023–2026

Key findings

• Theory · learning theory went from 86 papers in 2023 to 482 in 2026, roughly 5.6x in three years.
• NeurIPS publishes the most Theory · learning theory work: 540 papers from 2023 to 2026, 40% of the four-venue total.
• The fastest-growing overlap is with Social aspects · accountability transparency and interpretability: papers tagged with both rose from 0 in 2023 to 33 in 2026.
• Among the 5 topics under Theory and Statistics, it ranks #2 by 2026 paper count; the fastest grower in the group, Theory · optimization, is up 4.7x versus 5.6x here.

Representative papers

Spread across years and venues so every edition shows up

• NeurIPS 2026A dynamical systems theory of reward-modulated learning in linear recurrent networks
• ICML 2026A Fine-Grained Understanding of Uniform Convergence for Halfspaces
• ICLR 2026A Law of Data Reconstruction for Random Features (And Beyond)
• NeurIPS 2025A Geometric Analysis of PCA
• ICML 2025A Generalization Theory for Zero-Shot Prediction
• ICLR 2025A Solvable Attention for Neural Scaling Laws
• CVPR 2025Dynamic Stereotype Theory Induced Micro-expression Recognition with Oriented Deformation
• NeurIPS 20244+3 Phases of Compute-Optimal Neural Scaling Laws
• ICML 2024A Statistical Theory of Regularization-Based Continual Learning
• ICLR 2024A Sublinear Adversarial Training Algorithm
• CVPR 2024Accurate Training Data for Occupancy Map Prediction in Automated Driving Using Evidence Theory
• NeurIPS 2023$\varepsilon$-fractional core stability in Hedonic Games.
• ICML 2023Robust Collaborative Learning with Linear Gradient Overhead
• ICLR 2023Bispectral Neural Networks
• NeurIPS 2026A Mechanistic Investigation of Theory of Mind in a Large Language Model
• ICML 2026A Random Matrix Theory of Masked Self-Supervised Learning
• ICLR 2026A New Approach to Controlling Linear Dynamical Systems
• NeurIPS 2025Approximation theory for 1-Lipschitz ResNets
• ICML 2025A Two-Stage Learning-to-Defer Approach for Multi-Task Learning
• ICLR 2025A Theoretical Analysis of Self-Supervised Learning for Vision Transformers

All Theory · learning theory papers

540 papers, page 4 of 11

151. NeurIPS 2025Information-Theoretic Discrete Diffusion
152. NeurIPS 2025Infrequent Exploration in Linear Bandits
153. NeurIPS 2025Is Grokking a Computational Glass Relaxation?
154. NeurIPS 2025Knee-Deep in C-RASP: A Transformer Depth Hierarchy
155. NeurIPS 2025Last Iterate Convergence in Monotone Mean Field Games
156. NeurIPS 2025Learning from Interval Targets
157. NeurIPS 2025Learning Interestingness in Automated Mathematical Theory Formation
158. NeurIPS 2025Learning Linear Attention in Polynomial Time
159. NeurIPS 2025Learning Theory for Kernel Bilevel Optimization
160. NeurIPS 2025Length Generalization via Auxiliary Tasks
161. NeurIPS 2025Limitations of Normalization in Attention
162. NeurIPS 2025Low Precision Streaming PCA
163. NeurIPS 2025Machine Unlearning under Overparameterization
164. NeurIPS 2025Marginal-Nonuniform PAC Learnability
165. NeurIPS 2025Metric Automata Theory: A Unifying Theory of RNNs
166. NeurIPS 2025MindForge: Empowering Embodied Agents with Theory of Mind for Lifelong Cultural Learning
167. NeurIPS 2025Multidimensional Bayesian Utility Maximization: Tight Approximations to Welfare
168. NeurIPS 2025Neural Entropy
169. NeurIPS 2025Non-Asymptotic Analysis Of Data Augmentation For Precision Matrix Estimation
170. NeurIPS 2025Non-Stationary Lipschitz Bandits
171. NeurIPS 2025Non-Uniform Multiclass Learning with Bandit Feedback
172. NeurIPS 2025Nyström-Accelerated Primal LS-SVMs: Breaking the $O(an^3)$ Complexity Bottleneck for Scalable ODEs Learning
173. NeurIPS 2025On Agnostic PAC Learning in the Small Error Regime
174. NeurIPS 2025On Linear Mode Connectivity of Mixture-of-Experts Architectures
175. NeurIPS 2025On Minimax Estimation of Parameters in Softmax-Contaminated Mixture of Experts
176. NeurIPS 2025On the Emergence of Linear Analogies in Word Embeddings
177. NeurIPS 2025On the Optimality of the Median-of-Means Estimator under Adversarial Contamination
178. NeurIPS 2025On the Surprising Effectiveness of Large Learning Rates under Standard Width Scaling
179. NeurIPS 2025On the VC dimension of deep group convolutional neural networks
180. NeurIPS 2025On Transferring Transferability: Towards a Theory for Size Generalization
181. NeurIPS 2025On Universality Classes of Equivariant Networks
182. NeurIPS 2025Online Learning in the Repeated Mediated Newsvendor Problem
183. NeurIPS 2025Online Learning of Neural Networks
184. NeurIPS 2025Optimism Without Regularization: Constant Regret in Zero-Sum Games
185. NeurIPS 2025Pause Tokens Strictly Increase the Expressivity of Constant-Depth Transformers
186. NeurIPS 2025Posterior Sampling by Combining Diffusion Models with Annealed Langevin Dynamics
187. NeurIPS 2025Price of Parsimony: Complexity of Fourier Sparsity Testing
188. NeurIPS 2025Product Distribution Learning with Imperfect Advice
189. NeurIPS 2025Regularized least squares learning with heavy-tailed noise is minimax optimal
190. NeurIPS 2025Replicable Distribution Testing
191. NeurIPS 2025Revisiting Agnostic Boosting
192. NeurIPS 2025Revisiting Follow-the-Perturbed-Leader with Unbounded Perturbations in Bandit Problems
193. NeurIPS 2025Robust learning of halfspaces under log-concave marginals
194. NeurIPS 2025Robustly Learning Monotone Single-Index Models
195. NeurIPS 2025Safely Learning Controlled Stochastic Dynamics
196. NeurIPS 2025Sample-Conditional Coverage in Split-Conformal Prediction
197. NeurIPS 2025Self-Boost via Optimal Retraining: An Analysis via Approximate Message Passing
198. NeurIPS 2025Sharp Matrix Empirical Bernstein Inequalities
199. NeurIPS 2025Small Singular Values Matter: A Random Matrix Analysis of Transformer Models
200. NeurIPS 2025Smoothed Agnostic Learning of Halfspaces over the Hypercube

FAQ

How many Theory · learning theory papers were published at NeurIPS, ICLR, ICML and CVPR from 2023 to 2026?

1,350 in total: 86 in 2023, 393 in 2024, 389 in 2025, 482 in 2026.

How many Theory · learning theory papers are at NeurIPS 2026?

NeurIPS 2026 has 112 Theory · learning theory papers, versus 106 the year before.

How many Theory · learning theory papers are at ICLR 2026?

ICLR 2026 has 200 Theory · learning theory papers, versus 147 the year before.

How many Theory · learning theory papers are at ICML 2026?

ICML 2026 has 170 Theory · learning theory papers, versus 132 the year before.

How many Theory · learning theory papers are at CVPR 2025?

CVPR 2025 has 4 Theory · learning theory papers, versus 1 the year before.

Which conference publishes the most Theory · learning theory papers?

NeurIPS, with 540 papers from 2023 to 2026.