ICML 2026 accepted papers on Reinforcement learning
466 of the 6,341 papers accepted at ICML 2026 (7.3%) are on Reinforcement learning, compared with 6.5% at ICML 2025. They fall into 6 topics.
See ICML 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
RL from feedback · 168 papers
- How RLHF Amplifies Sycophancy (2 citations)
- Human-in-the-Loop Policy Optimization for Preference-Based Multi-Objective Reinforcement Learning (2 citations)
- OmniSapiens: A Foundation Model for Social Behavior Processing via Heterogeneity-Aware Relative Policy Optimization (2 citations)
- Effects of Structural Reward Shaping on Biophysical Properties in RL-Trained Plasmid Generators (1 citation)
- Advantage Collapse in Group Relative Policy Optimization: Diagnosis and Mitigation
- Advantage Weighted Matching: Aligning RL with Pretraining in Diffusion Models
- Agentic Monte Carlo: Simulating Reinforcement Learning for Black-Box Agents
- Alternating Reinforcement Learning for Rubric-Based Reward Modeling in Non-Verifiable LLM Post-Training
- Anchored Policy Optimization: Mitigating Exploration Collapse via Support-Constrained Rectification
- Approximation of Log-Partition Function in Policy Mirror Descent Induces Implicit Regularization for LLM Post-Training
- AReaL-DTA: Dynamic Tree Attention for Efficient Reinforcement Learning of Large Language Models
- ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking
and 156 more in the interactive map.
Policy gradient and actor-critic · 110 papers
- MiniMax Learning of Interpretable Factored Stochastic Policies from Conjoint Data, with Uncertainty Quantification (1 citation)
- $V_0$: A Generalist Value Model for Any Policy at State Zero
- A KL-regularization framework for learning to plan with adaptive priors
- AC-ODM: Actor–Critic Online Data Mixing for Sample-Efficient LLM Pretraining
- Accelerating Q-learning through Efficient Value-Sharing across Actions
- Beyond the Bellman Recursion: A Pontryagin-Guided Framework for Non-Exponential Discounting
- Boosting CVaR Policy Optimization with Quantile Gradients
- Bridging the Gap Between Average and Discounted TD Learning
- Chunk-Guided Q-Learning
- Conformal Policy Control
- Constrained Multi-Objective Reinforcement Learning with Max-Min Criterion
- Convergence of Two-Timescale Markovian Stochastic Approximations with Applications in Reinforcement Learning
and 98 more in the interactive map.
Visual and representation RL · 100 papers
- Action-Sufficient Goal Representations
- Adaptive Quasimetric Mapping : Principled Topological Abstraction for Robust Offline Goal-Conditioned Navigation
- Adaptive Reinforcement Learning for Unobservable Random Delays
- Alignment Risks from Capability-Seeking RL Training
- ASAP: Exploiting the Satisficing Generalization Edge in Neural Combinatorial Optimization
- Avoid What You Know: Divergent Trajectory Balance for GFlowNets
- Cerebellar-Inspired Residual Control for Fault Recovery: From Inference-Time Adaptation to Structural Consolidation
- Chain-of-Goals Hierarchical Policy for Long-Horizon Offline Goal-Conditioned RL
- Chebyshev Policies and the Mountain Car Problem: Reinforcement Learning for Low-Dimensional Control Tasks
- Compositional Planning with Jumpy World Models
- Compositional Transduction with Latent Analogies for Offline Goal-Conditioned Reinforcement Learning
- CorrectionPlanner: Self-Correction Planner with Reinforcement Learning in Autonomous Driving
and 88 more in the interactive map.
Offline RL · 39 papers
- AdamO: A Collapse-Suppressed Optimizer for Offline RL
- Behavior-Invariant Task Representation Learning with Transformer-based World Models for Offline Meta-Reinforcement Learning
- Beyond the Proxy: Trajectory-Distilled Guidance for Offline GFlowNet Training
- BiTrajDiff: Bidirectional Trajectory Generation with Diffusion Models for Offline Reinforcement Learning
- Causal Flow Q-Learning for Robust Offline Reinforcement Learning
- DRIVE: Distributional and Retrieval-Augmented Bidding with Value Evaluation
- Efficient and Uncertainty-Aware Diffusion Framework for Offline-to-Online Reinforcement Learning
- Fast and Highly Expressive Policy Learning for Offline Reinforcement Learning via Bootstrapped Flow Q-Learning
- From Static Constraints to Dynamic Adaptation: Sample-Level Constraint Relaxation for Offline-to-Online Reinforcement Learning
- Improving Diffusion Planners by Self-Supervised Action Gating with Energies
- Improving Zero-Shot Offline RL via Behavioral Task Sampling
- Long-Horizon Model-Based Offline Reinforcement Learning Without Explicit Conservatism
and 27 more in the interactive map.
MDPs and planning · 29 papers
- Bilevel Optimization over Saddle Points of Zero-Sum Markov Games
- Breaking the Computational Barrier: Provably Efficient Actor–Critic for Low-Rank MDPs
- Commit to the Bit: Reactive Reinforcement Learning Done Right
- Constrained Meta Reinforcement Learning with Provable Test-Time Safety
- Dynamic Programming for Epistemic Uncertainty in Markov Decision Processes
- Fast Non-Episodic Finite-Horizon RL with K-Step Lookahead Thresholding
- From Generative to Episodic: Sample-Efficient Replicable Reinforcement Learning
- Impact of Connectivity on Laplacian Representations in Reinforcement Learning
- Improved Bounds for Reward-Agnostic and Reward-Free Exploration
- Learn to change the world: Multi-level reinforcement learning with model-changing actions
- Learning the Minimum Action Distance
- Model-Free Robust Average-Reward Reinforcement Learning with Sample Complexity Analysis
and 17 more in the interactive map.
Imitation learning · 20 papers
- Adversarial Dual On-Policy Distillation from Expressive Teacher
- Consistent Zero-Shot Imitation with Contrastive Goal Inference
- Distinguishing Imitation Error from Intrinsic Motion Learning Difficulty
- Distributional Inverse Reinforcement Learning
- Escaping the Verifier: Learning to Reason via Demonstrations
- FAIL: Flow Matching Adversarial Imitation Learning for Image Generation
- NavOL: Navigation Policy with Online Imitation Learning
- Non-Adversarial Imitation Learning Provably Free of Compounding Errors: The Value Flow Mechanism
- On the Sample Efficiency of Inverse Dynamics Models for Semi-Supervised Imitation Learning
- Posterior Behavioral Cloning: Pretraining BC Policies for Efficient RL Finetuning
- Provably Efficient Policy-Reward Co-Pretraining for Adversarial Imitation Learning
- Recovering Hidden Reward in Diffusion-Based Policies
and 8 more in the interactive map.