Theory · online learning papers, 2023–2026

Key findings

• Theory · online learning went from 164 papers in 2023 to 190 in 2026, roughly 1.2x in three years.
• NeurIPS publishes the most Theory · online learning work: 427 papers from 2023 to 2026, 56% of the four-venue total.
• The fastest-growing overlap is with Simulation for Driving: papers tagged with both rose from 1 in 2023 to 12 in 2026.
• Among the 5 topics under Theory and Statistics, it ranks #3 by 2026 paper count; the fastest grower in the group, Theory · learning theory, is up 5.6x versus 1.2x here.

Representative papers

Spread across years and venues so every edition shows up

• NeurIPS 2026A Single-Sample Polylogarithmic Regret Bound for Nonstationary Online Linear Programming
• ICML 2026A Perturbation Approach to Unconstrained Linear Bandits
• ICLR 2026Rethinking Data Curation in LLM Training: Online Reweighting Offers Better Generalization than Offline Methods
• CVPR 2026An Optimal Transport-driven Approach for Cultivating Latent Space in Online
• NeurIPS 2025A Learning-Augmented Approach to Online Allocation Problems
• ICML 2025A Classification View on Meta Learning Bandits
• ICLR 2025Efficient Online Pruning and Abstraction for Imperfect Information Extensive-Form Games
• CVPR 2025InteractionMap: Improving Online Vectorized HDMap Construction with Interaction
• NeurIPS 2024A Best-of-both-worlds Algorithm for Bandits with Delayed Feedback with Robustness to Excessive Delays
• ICML 2024On Online Experimentation without Device Identifiers
• ICLR 2024Improved Regret Bounds for Non-Convex Online-Within-Online Meta Learning
• CVPR 2024Dual-Enhanced Coreset Selection with Class-wise Collaboration for Online Blurry Class Incremental Learning
• NeurIPS 2023$\varepsilon$-fractional core stability in Hedonic Games.
• ICML 2023Applied Online Algorithms with Heterogeneous Predictors
• ICLR 2023Hebbian Deep Learning Without Feedback
• CVPR 2023VisFusion: Visibility-Aware Online 3D Scene Reconstruction From Videos
• NeurIPS 2026A Theory of Adversary-Directed Online Learning
• ICML 2026A Two-Layer Framework for Joint Online Configuration Selection and Admission Control
• ICLR 2026A Statistical Learning Perspective on Semi-dual Adversarial Neural Optimal Transport Solvers
• CVPR 2026Dynamic Momentum Recalibration in Online Gradient Learning

All Theory · online learning papers

763 papers, page 3 of 16

101. ICML 2026Optimal Anytime Algorithms for Online Convex Optimization with Adversarial Constraints
102. ICML 2026Optimal Bayesian Stopping for Efficient Inference of Consistent LLM Answers
103. ICML 2026Parsimonious Learning-Augmented Online Metric Matching
104. ICML 2026Prediction-Powered Adaptive Inference with Pretrained AI Models for Contextual Bandits
105. ICML 2026Pushing the Limits of Block Rotations in Post-Training Quantization
106. ICML 2026Regret Minimization With a Crowd of Awakening Experts
107. ICML 2026Revenue Guarantees of No-Swap-Regret Dynamics in First Price Auctions
108. ICML 2026Robust and Consistent Ski Rental with Distributional Advice
109. ICML 2026SCORE: A Unified Framework for Overshoot Refund in Online FDR Control
110. ICML 2026Ski Rental with Distributional Predictions of Unknown Quality
111. ICML 2026SMAC: Score-Matched Actor-Critics for Robust Offline-to-Online Transfer
112. ICML 2026Stochastic Linear Bandits with Parameter Noise
113. ICML 2026Stronger Benchmarks for Prediction as a Service with Constraints
114. ICML 2026The Cost of Learning Under Multiple Change Points
115. ICML 2026Theory of Continual Learning Against Data Poisoning Attacks
116. ICML 2026Tighter Regret Lower Bound for Gaussian Process Bandits with Squared Exponential Kernel in Hypersphere
117. ICML 2026Universal Multiclass Transductive Online Learning
118. ICML 2026Upper-Linearizability of Online Non-Monotone DR-Submodular Maximization over Down-Closed Convex Sets
119. ICML 2026What Does Flow-Matching Bring to TD-Learning?
120. ICML 2026What Does Thompson Sampling Optimize?
121. ICML 2026When Drafts Evolve: Speculative Decoding Meets Online Learning
122. ICML 2026When Softmax Fails at the Top: Extreme‑Value Corrections for InfoNCE
123. NeurIPS 2026A Compass for Useful Data: Online Data Selection via Alignment-Gated Fisher Geometry
124. NeurIPS 2026A Single-Sample Polylogarithmic Regret Bound for Nonstationary Online Linear Programming
125. NeurIPS 2026A Theory of Adversary-Directed Online Learning
126. NeurIPS 2026A Theory of Online Learning with Autoregressive Chain-of-Thought Reasoning
127. NeurIPS 2026Adaptive Calibration in Non-Stationary Environments
128. NeurIPS 2026Agnostic Online Learning with Reliable Abstention
129. NeurIPS 2026BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates
130. NeurIPS 2026Beyond Row Alignment: Virtual-Camera-Aware Online Stereo Rectification
131. NeurIPS 2026Bian Que: An Agentic Framework with Flexible Skill Arrangement for Online System Operations
132. NeurIPS 2026Bilinear Matching Bandits
133. NeurIPS 2026Curvature-Dependent Lower Bounds for Riemannian Online Convex Optimization
134. NeurIPS 2026Deployment-Time Online Imitation Learning from Corrective Demonstrations
135. NeurIPS 2026Direction-Aware Offline-to-Online Learning in Linear Contextual Bandits
136. NeurIPS 2026Evidence Guided Dual Expert Memory with Joint Routing for VLLMs Online Correction
137. NeurIPS 2026Filtered-Trace Online Variational Training for Probabilistic Spiking Neural Networks
138. NeurIPS 2026First-Order Regret for Online Convex Optimization with Memory and Online Nonstochastic Control
139. NeurIPS 2026Flux: Online, Fine-Grained Data Scheduling for Training Machine Learning Interatomic Potentials
140. NeurIPS 2026From Generic to Dedicated: A Novel Optimizer for Online Continual Learning
141. NeurIPS 2026From POMDP Theory to Deep RL with Particle Filters
142. NeurIPS 2026Geometry-Aware Online Scheduling for LLM Serving: From Theoretical Bound to System Practice
143. NeurIPS 2026GHOST: Geometry-Hierarchical Online Streaming Token Eviction for Efficient 3D Reconstruction
144. NeurIPS 2026HALO-VGGT: Heterogeneity-Aware Lightweight Online Compression Allocator for Efficient VGGT
145. NeurIPS 2026High Probability Risk Control for Online Policy Learning
146. NeurIPS 2026Improved Algorithms for Online Classification with Surrogate Losses
147. NeurIPS 2026Instance-Adaptive Online Multicalibration
148. NeurIPS 2026Learning to Price with Persuasion
149. NeurIPS 2026Learning-Augmented Online Portfolio Selection: Optimal Robustness-Consistency Tradeoffs
150. NeurIPS 2026Learning-Augmented Online Scheduling with Parsimonious Preemption

FAQ

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

763 in total: 164 in 2023, 224 in 2024, 185 in 2025, 190 in 2026.

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

NeurIPS 2026 has 68 Theory · online learning papers, versus 89 the year before.

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

ICLR 2026 has 27 Theory · online learning papers, versus 18 the year before.

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

ICML 2026 has 86 Theory · online learning papers, versus 74 the year before.

How many Theory · online learning papers are at CVPR 2026?

CVPR 2026 has 9 Theory · online learning papers, versus 4 the year before.

Which conference publishes the most Theory · online learning papers?

NeurIPS, with 427 papers from 2023 to 2026.