ICML 2023 accepted papers on Reinforcement learning

Visual and representation RL · 63 papers

• CleanRL: High-quality Single-file Implementations of Deep Reinforcement Learning Algorithms (86 citations)
• Grounding Large Language Models in Interactive Environments with Online Reinforcement Learning (35 citations)
• Human-Timescale Adaptation in an Open-Ended Task Space (20 citations)
• Hyperparameters in Reinforcement Learning and How To Tune Them (9 citations)
• Controllability-Aware Unsupervised Skill Discovery (8 citations)
• An Investigation into Pre-Training Object-Centric Representations for Reinforcement Learning (7 citations)
• The Dormant Neuron Phenomenon in Deep Reinforcement Learning (7 citations)
• Understanding Plasticity in Neural Networks (7 citations)
• Bigger, Better, Faster: Human-level Atari with human-level efficiency (5 citations)
• Graph Reinforcement Learning for Network Control via Bi-Level Optimization (5 citations)
• Mastering the Unsupervised Reinforcement Learning Benchmark from Pixels (5 citations)
• Retrosynthetic Planning with Dual Value Networks (5 citations)

and 51 more in the interactive map.

Policy gradient and actor-critic · 51 papers

• Policy Evaluation and Temporal-Difference Learning in Continuous Time and Space: A Martingale Approach (38 citations)
• Enforcing Hard Constraints with Soft Barriers: Safe Reinforcement Learning in Unknown Stochastic Environments (8 citations)
• Quantum Policy Gradient Algorithm with Optimized Action Decoding (6 citations)
• ReLOAD: Reinforcement Learning with Optimistic Ascent-Descent for Last-Iterate Convergence in Constrained MDPs (5 citations)
• The Benefits of Model-Based Generalization in Reinforcement Learning (3 citations)
• Better Training of GFlowNets with Local Credit and Incomplete Trajectories (2 citations)
• Live in the Moment: Learning Dynamics Model Adapted to Evolving Policy (2 citations)
• On the Convergence of SARSA with Linear Function Approximation (2 citations)
• Optimal Goal-Reaching Reinforcement Learning via Quasimetric Learning (2 citations)
• Posterior Sampling for Deep Reinforcement Learning (2 citations)
• Eventual Discounting Temporal Logic Counterfactual Experience Replay (1 citation)
• Flipping Coins to Estimate Pseudocounts for Exploration in Reinforcement Learning (1 citation)

and 39 more in the interactive map.

Offline RL · 34 papers

• Efficient Online Reinforcement Learning with Offline Data (13 citations)
• Q-learning Decision Transformer: Leveraging Dynamic Programming for Conditional Sequence Modelling in Offline RL (12 citations)
• ChiPFormer: Transferable Chip Placement via Offline Decision Transformer (10 citations)
• Constrained Decision Transformer for Offline Safe Reinforcement Learning (8 citations)
• Contrastive Energy Prediction for Exact Energy-Guided Diffusion Sampling in Offline Reinforcement Learning (3 citations)
• Distance Weighted Supervised Learning for Offline Interaction Data (3 citations)
• Leveraging Offline Data in Online Reinforcement Learning (3 citations)
• Policy Regularization with Dataset Constraint for Offline Reinforcement Learning (3 citations)
• Safe Offline Reinforcement Learning with Real-Time Budget Constraints (3 citations)
• Actor-Critic Alignment for Offline-to-Online Reinforcement Learning (2 citations)
• Future-conditioned Unsupervised Pretraining for Decision Transformer (2 citations)
• Masked Trajectory Models for Prediction, Representation, and Control (2 citations)

and 22 more in the interactive map.

MDPs and planning · 28 papers

• Hindsight Learning for MDPs with Exogenous Inputs (4 citations)
• Nearly Minimax Optimal Reinforcement Learning for Linear Markov Decision Processes (4 citations)
• Model-Free Robust Average-Reward Reinforcement Learning (3 citations)
• Identifiability and Generalizability in Constrained Inverse Reinforcement Learning (2 citations)
• Policy Gradient in Robust MDPs with Global Convergence Guarantee (2 citations)
• Non-stationary Reinforcement Learning under General Function Approximation (1 citation)
• Predictable MDP Abstraction for Unsupervised Model-Based RL (1 citation)
• The Unintended Consequences of Discount Regularization: Improving Regularization in Certainty Equivalence Reinforcement Learning (1 citation)
• Computationally Efficient PAC RL in POMDPs with Latent Determinism and Conditional Embeddings
• Fast Rates for Maximum Entropy Exploration
• Low-Switching Policy Gradient with Exploration via Online Sensitivity Sampling
• Multi-User Reinforcement Learning with Low Rank Rewards

and 16 more in the interactive map.

Imitation learning · 11 papers

• Jump-Start Reinforcement Learning (17 citations)
• Hierarchical Imitation Learning with Vector Quantized Models (3 citations)
• Invariance in Policy Optimisation and Partial Identifiability in Reward Learning (3 citations)
• Towards Theoretical Understanding of Inverse Reinforcement Learning (3 citations)
• A Coupled Flow Approach to Imitation Learning (2 citations)
• Inverse Reinforcement Learning without Reinforcement Learning (1 citation)
• SeMAIL: Eliminating Distractors in Visual Imitation via Separated Models (1 citation)
• Active Policy Improvement from Multiple Black-box Oracles
• Learning Control by Iterative Inversion
• Multi-task Hierarchical Adversarial Inverse Reinforcement Learning
• Policy Contrastive Imitation Learning

RL from feedback · 4 papers

• Scaling Laws for Reward Model Overoptimization (34 citations)
• Principled Reinforcement Learning with Human Feedback from Pairwise or K-wise Comparisons (16 citations)
• Aligning Language Models with Preferences through $f$-divergence Minimization (5 citations)
• Optimizing DDPM Sampling with Shortcut Fine-Tuning (4 citations)