Theory · optimization papers, 2023–2026

Key findings

• Theory · optimization went from 123 papers in 2023 to 584 in 2026, roughly 4.7x in three years.
• NeurIPS publishes the most Theory · optimization work: 778 papers from 2023 to 2026, 42% of the four-venue total.
• The fastest-growing overlap is with Low-Rank Adaptation: papers tagged with both rose from 0 in 2023 to 31 in 2026.
• Among the 5 topics under Theory and Statistics, it ranks #1 by 2026 paper count; the fastest grower in the group, Theory · learning theory, is up 5.6x versus 4.7x here.

Representative papers

Spread across years and venues so every edition shows up

• NeurIPS 2026A Mechanistic Investigation of Theory of Mind in a Large Language Model
• ICML 2026A Direct Second-Order Method for Solving Two-Player Zero-Sum Games
• ICLR 2026A Convergence Analysis of Adaptive Optimizers under Floating-point Quantization
• CVPR 2026TriSim Tri-Dimensional Similarity Modeling with Extreme Value Theory for False-Negative
• NeurIPS 20253BASiL: An Algorithmic Framework for Sparse plus Low-Rank Compression of LLMs
• ICML 2025Active Learning of Deep Neural Networks via Gradient-Free Cutting Planes
• ICLR 2025A Coefficient Makes SVRG Effective
• CVPR 2025Dynamic Stereotype Theory Induced Micro-expression Recognition with Oriented Deformation
• NeurIPS 2024A Combinatorial Algorithm for the Semi-Discrete Optimal Transport Problem
• ICML 2024A Statistical Theory of Regularization-Based Continual Learning
• ICLR 2024A Discretization Framework for Robust Contextual Stochastic Optimization
• NeurIPS 2023[Re] Bandit Theory and Thompson Sampling-guided Directed Evolution for Sequence Optimization
• ICML 2023Averaged Method of Multipliers for Bi-Level Optimization without Lower-Level Strong Convexity
• ICLR 2023Adaptive Optimization in the $\infty$-Width Limit
• CVPR 2023Deep Hashing With Minimal-Distance-Separated Hash Centers
• NeurIPS 2026Avoiding Obfuscation with Prover-Estimator Debate
• ICML 2026A General Framework for Dynamic Consistent Submodular Maximization
• ICLR 2026A Memory-Efficient Hierarchical Algorithm for Large-scale Optimal Transport Problems
• CVPR 2026FedAdamom Adaptive Momentum for Improved Generalization in Federated Optimization
• NeurIPS 2025A Beyond-Worst-Case Analysis of Greedy k-means++

All Theory · optimization papers

778 papers, page 3 of 16

101. NeurIPS 2026Sharp Non-Asymptotic Analysis of the Penalized Challenger in $\beta$-EB-TCI for Bernoulli Bandits
102. NeurIPS 2026Sign-Based Optimizers Are Effective Under Heavy-Tailed Noise
103. NeurIPS 2026Skip the Hessian, Keep the Rates: Globalized Semismooth Newton with Lazy Hessian Updates
104. NeurIPS 2026Stochastic Zeroth-Order Optimization Under Heavy-Tailed Noise
105. NeurIPS 2026Structure vs. Chaos: Asymmetric Entropic Optimization for Enforcing Instruction Hierarchy
106. NeurIPS 2026Subcritical Signal Propagation at Initialization in Normalization-Free Transformers
107. NeurIPS 2026Taming the Tails: Why Distributionally Robust Optimization Needs New Theory for Imbalanced Regression
108. NeurIPS 2026The Optimization Prior: Instilling depth for shallow networks, detail for coarse networks
109. NeurIPS 2026The Symmetries of Three-Layer ReLU Networks
110. NeurIPS 2026Theory on Attention Dynamics for Out-of-Distribution In-Context Learning
111. NeurIPS 2026Tight Generalization Bounds for Noiseless Inverse Optimization
112. NeurIPS 2026Toward a theory of Evaluability
113. NeurIPS 2026Towards Settling the Complexity of Non-Euclidean Parallel Convex Optimization
114. NeurIPS 2026Training-free Spatially Grounded Geometric Shape Encoding
115. NeurIPS 2026Training-Induced Escape from Token Clustering in a Mean-Field Formulation of Transformers
116. NeurIPS 2026Truncated Riemannian (1+1)-ES for Black-Box Optimization with Intrinsic Dimension Guarantees
117. NeurIPS 2026Value-Priced Uncertainty: A Family of PPO-Compatible Exploration Bonuses
118. NeurIPS 2026VSPO: Vector-Steered Policy Optimization for Controllable Model Behavior
119. NeurIPS 2026When Symbol Names Should Not Matter: A Logistic Theory of Fresh-Symbol Classification
120. NeurIPS 20253BASiL: An Algorithmic Framework for Sparse plus Low-Rank Compression of LLMs
121. NeurIPS 2025A Beyond-Worst-Case Analysis of Greedy k-means++
122. NeurIPS 2025A Computationally Viable Numerical Gradient-based Technique for Optimal Covering Problems
123. NeurIPS 2025A Difference-of-Convex Functions Approach to Energy-Based Iterative Reasoning
124. NeurIPS 2025A Geometric Analysis of PCA
125. NeurIPS 2025A Gradient Guided Diffusion Framework for Chance Constrained Programming
126. NeurIPS 2025A Learning-Augmented Dynamic Programming Approach for Orienteering Problem with Time Windows
127. NeurIPS 2025A Minimalist Example of Edge-of-Stability and Progressive Sharpening
128. NeurIPS 2025A Near-Optimal Algorithm for Decentralized Convex-Concave Finite-Sum Minimax Optimization
129. NeurIPS 2025A Regularized Newton Method for Nonconvex Optimization with Global and Local Complexity Guarantees
130. NeurIPS 2025A Single-Loop First-Order Algorithm for Linearly Constrained Bilevel Optimization
131. NeurIPS 2025A Single-Loop Gradient Algorithm for Pessimistic Bilevel Optimization via Smooth Approximation
132. NeurIPS 2025A Statistical Theory of Contrastive Learning via Approximate Sufficient Statistics
133. NeurIPS 2025A Tale of Two Symmetries: Exploring the Loss Landscape of Equivariant Models
134. NeurIPS 2025A Theoretical Framework for Grokking: Interpolation followed by Riemannian Norm Minimisation
135. NeurIPS 2025A Theory for Worst-Case vs. Average-Case Guarantees for LLMs
136. NeurIPS 2025A Unified Analysis of Stochastic Gradient Descent with Arbitrary Data Permutations and Beyond
137. NeurIPS 2025A Unified Stability Analysis of SAM vs SGD: Role of Data Coherence and Emergence of Simplicity Bias
138. NeurIPS 2025Accelerated Distance-adaptive Methods for Hölder Smooth and Convex Optimization
139. NeurIPS 2025Accelerating Block Coordinate Descent for LLM Finetuning via Landscape Expansion
140. NeurIPS 2025Accelerating Model-Free Optimization via Averaging of Cost Samples
141. NeurIPS 2025Accelerating Optimization via Differentiable Stopping Time
142. NeurIPS 2025Acceleration via silver step-size on Riemannian manifolds with applications to Wasserstein space
143. NeurIPS 2025Adam Reduces a Unique Form of Sharpness: Theoretical Insights Near the Minimizer Manifold
144. NeurIPS 2025Adaptive Algorithms with Sharp Convergence Rates for Stochastic Hierarchical Optimization
145. NeurIPS 2025Adaptive Frontier Exploration on Graphs with Applications to Network-Based Disease Testing
146. NeurIPS 2025Adaptive Riemannian ADMM for Nonsmooth Optimization: Optimal Complexity without Smoothing
147. NeurIPS 2025Affine-Invariant Global Non-Asymptotic Convergence Analysis of BFGS under Self-Concordance
148. NeurIPS 2025Agnostic Active Learning Is Always Better Than Passive Learning
149. NeurIPS 2025Alternating Gradient Flows: A Theory of Feature Learning in Two-layer Neural Networks
150. NeurIPS 2025An Adaptive Algorithm for Bilevel Optimization on Riemannian Manifolds

FAQ

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

1,871 in total: 123 in 2023, 494 in 2024, 670 in 2025, 584 in 2026.

How many Theory · optimization papers are at NeurIPS 2026?

NeurIPS 2026 has 119 Theory · optimization papers, versus 318 the year before.

How many Theory · optimization papers are at ICLR 2026?

ICLR 2026 has 244 Theory · optimization papers, versus 196 the year before.

How many Theory · optimization papers are at ICML 2026?

ICML 2026 has 214 Theory · optimization papers, versus 147 the year before.

How many Theory · optimization papers are at CVPR 2026?

CVPR 2026 has 7 Theory · optimization papers, versus 9 the year before.

Which conference publishes the most Theory · optimization papers?

NeurIPS, with 778 papers from 2023 to 2026.