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

389 papers, page 2 of 8

51. ICLR 2025From Sparse Dependence to Sparse Attention: Unveiling How Chain-of-Thought Enhances Transformer Sample Efficiency
52. ICLR 2025Fundamental Limitations on Subquadratic Alternatives to Transformers
53. ICLR 2025Generalizability of Neural Networks Minimizing Empirical Risk Based on Expressive Power
54. ICLR 2025Generalization Bounds and Model Complexity for Kolmogorov–Arnold Networks
55. ICLR 2025Generalization Bounds for Canonicalization: A Comparative Study with Group Averaging
56. ICLR 2025Generalization Guarantees for Representation Learning via Data-Dependent Gaussian Mixture Priors
57. ICLR 2025Generalized Principal-Agent Problem with a Learning Agent
58. ICLR 2025Geometry of Lightning Self-Attention: Identifiability and Dimension
59. ICLR 2025Global Convergence in Neural ODEs: Impact of Activation Functions
60. ICLR 2025Grokking at the Edge of Numerical Stability
61. ICLR 2025How DNNs break the Curse of Dimensionality: Compositionality and Symmetry Learning
62. ICLR 2025How Feature Learning Can Improve Neural Scaling Laws
63. ICLR 2025How Gradient descent balances features: A dynamical analysis for two-layer neural networks
64. ICLR 2025How many samples are needed to train a deep neural network?
65. ICLR 2025How Much is Unseen Depends Chiefly on Information About the Seen
66. ICLR 2025Improved Algorithms for Kernel Matrix-Vector Multiplication Under Sparsity Assumptions
67. ICLR 2025Improved Convergence Rate for Diffusion Probabilistic Models
68. ICLR 2025Improved Finite-Particle Convergence Rates for Stein Variational Gradient Descent
69. ICLR 2025Inner Information Analysis Algorithm for Deep Neural Network based on Community
70. ICLR 2025Lambda-Skip Connections: the architectural component that prevents Rank Collapse
71. ICLR 2025Lasso Bandit with Compatibility Condition on Optimal Arm
72. ICLR 2025Last-Iterate Convergence Properties of Regret-Matching Algorithms in Games
73. ICLR 2025Learning from weak labelers as constraints
74. ICLR 2025Learning Hierarchical Polynomials of Multiple Nonlinear Features
75. ICLR 2025Learning High-Degree Parities: The Crucial Role of the Initialization
76. ICLR 2025Learning Multi-Index Models with Neural Networks via Mean-Field Langevin Dynamics
77. ICLR 2025Learning Neural Networks with Distribution Shift: Efficiently Certifiable Guarantees
78. ICLR 2025Learning to Help in Multi-Class Settings
79. ICLR 2025Learning-Augmented Search Data Structures
80. ICLR 2025Let Me Grok for You: Accelerating Grokking via Embedding Transfer from a Weaker Model
81. ICLR 2025Leveraging Flatness to Improve Information-Theoretic Generalization Bounds for SGD
82. ICLR 2025Limits of Deep Learning: Sequence Modeling through the Lens of Complexity Theory
83. ICLR 2025Limits to scalable evaluation at the frontier: LLM as judge won’t beat twice the data
84. ICLR 2025Lipschitz Bandits in Optimal Space
85. ICLR 2025Make Haste Slowly: A Theory of Emergent Structured Mixed Selectivity in Feature Learning ReLU Networks
86. ICLR 2025MAST: model-agnostic sparsified training
87. ICLR 2025Maximizing the Potential of Synthetic Data: Insights from Random Matrix Theory
88. ICLR 2025Minimax Optimal Two-Stage Algorithm For Moment Estimation Under Covariate Shift
89. ICLR 2025Misspecified $Q$-Learning with Sparse Linear Function Approximation: Tight Bounds on Approximation Error
90. ICLR 2025Near-optimal Active Regression of Single-Index Models
91. ICLR 2025No Free Lunch: Fundamental Limits of Learning Non-Hallucinating Generative Models
92. ICLR 2025O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions
93. ICLR 2025On the Convergence of No-Regret Dynamics in Information Retrieval Games with Proportional Ranking Functions
94. ICLR 2025On the expressiveness and spectral bias of KANs
95. ICLR 2025On the Expressiveness of Rational ReLU Neural Networks With Bounded Depth
96. ICLR 2025On the Learn-to-Optimize Capabilities of Transformers in In-Context Sparse Recovery
97. ICLR 2025On the Optimal Memorization Capacity of Transformers
98. ICLR 2025On the Optimization and Generalization of Two-layer Transformers with Sign Gradient Descent
99. ICLR 2025On the Relation between Trainability and Dequantization of Variational Quantum Learning Models
100. ICLR 2025ONLINE EPSILON NET & PIERCING SET FOR GEOMETRIC CONCEPTS

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.