ICLR 2023 accepted papers on Reinforcement learning

Visual and representation RL · 58 papers

• VIP: Towards Universal Visual Reward and Representation via Value-Implicit Pre-Training (35 citations)
• ERL-Re$^2$: Efficient Evolutionary Reinforcement Learning with Shared State Representation and Individual Policy Representation (16 citations)
• In-context Reinforcement Learning with Algorithm Distillation (10 citations)
• PD-MORL: Preference-Driven Multi-Objective Reinforcement Learning Algorithm (9 citations)
• MoDem: Accelerating Visual Model-Based Reinforcement Learning with Demonstrations (8 citations)
• DEP-RL: Embodied Exploration for Reinforcement Learning in Overactuated and Musculoskeletal Systems (6 citations)
• Fantastic Rewards and How to Tame Them: A Case Study on Reward Learning for Task-oriented Dialogue Systems (6 citations)
• From Play to Policy: Conditional Behavior Generation from Uncurated Robot Data (6 citations)
• Neural Episodic Control with State Abstraction (5 citations)
• Neuroevolution is a Competitive Alternative to Reinforcement Learning for Skill Discovery (5 citations)
• POPGym: Benchmarking Partially Observable Reinforcement Learning (5 citations)
• Choreographer: Learning and Adapting Skills in Imagination (4 citations)

and 46 more in the interactive map.

Offline RL · 29 papers

• Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learning (34 citations)
• Is Conditional Generative Modeling all you need for Decision Making? (31 citations)
• Benchmarking Offline Reinforcement Learning on Real-Robot Hardware (11 citations)
• Behavior Proximal Policy Optimization (8 citations)
• Offline RL for Natural Language Generation with Implicit Language Q Learning (8 citations)
• Hybrid RL: Using both offline and online data can make RL efficient (7 citations)
• Offline Reinforcement Learning via High-Fidelity Generative Behavior Modeling (6 citations)
• Extreme Q-Learning: MaxEnt RL without Entropy (5 citations)
• Policy Expansion for Bridging Offline-to-Online Reinforcement Learning (5 citations)
• Offline RL with No OOD Actions: In-Sample Learning via Implicit Value Regularization (4 citations)
• When Data Geometry Meets Deep Function: Generalizing Offline Reinforcement Learning (4 citations)
• Behavior Prior Representation learning for Offline Reinforcement Learning (3 citations)

and 17 more in the interactive map.

Policy gradient and actor-critic · 27 papers

• Deep Reinforcement Learning for Cost-Effective Medical Diagnosis (14 citations)
• Efficient Deep Reinforcement Learning Requires Regulating Overfitting (9 citations)
• On the Robustness of Safe Reinforcement Learning under Observational Perturbations (9 citations)
• CFlowNets: Continuous Control with Generative Flow Networks (5 citations)
• Guarded Policy Optimization with Imperfect Online Demonstrations (4 citations)
• Decision S4: Efficient Sequence-Based RL via State Spaces Layers (3 citations)
• Safe Reinforcement Learning From Pixels Using a Stochastic Latent Representation (3 citations)
• A CMDP-within-online framework for Meta-Safe Reinforcement Learning (2 citations)
• Provably Efficient Risk-Sensitive Reinforcement Learning: Iterated CVaR and Worst Path (2 citations)
• Timing is Everything: Learning to Act Selectively with Costly Actions and Budgetary Constraints (2 citations)
• Latent State Marginalization as a Low-cost Approach for Improving Exploration (1 citation)
• Quasi-optimal Reinforcement Learning with Continuous Actions (1 citation)

and 15 more in the interactive map.

MDPs and planning · 19 papers

• Partially Observable RL with B-Stability: Unified Structural Condition and Sharp Sample-Efficient Algorithms (2 citations)
• Variational Latent Branching Model for Off-Policy Evaluation (2 citations)
• VIPeR: Provably Efficient Algorithm for Offline RL with Neural Function Approximation (2 citations)
• A General Framework for Sample-Efficient Function Approximation in Reinforcement Learning (1 citation)
• PAC Reinforcement Learning for Predictive State Representations (1 citation)
• Sample Complexity of Nonparametric Off-Policy Evaluation on Low-Dimensional Manifolds using Deep Networks (1 citation)
• Improved Sample Complexity for Reward-free Reinforcement Learning under Low-rank MDPs
• Near-Optimal Deployment Efficiency in Reward-Free Reinforcement Learning with Linear Function Approximation
• Near-optimal Policy Identification in Active Reinforcement Learning
• Pessimism in the Face of Confounders: Provably Efficient Offline Reinforcement Learning in Partially Observable Markov Decision Processes
• Safe Exploration Incurs Nearly No Additional Sample Complexity for Reward-Free RL
• Wasserstein Auto-encoded MDPs: Formal Verification of Efficiently Distilled RL Policies with Many-sided Guarantees

and 7 more in the interactive map.

Imitation learning · 11 papers

• CLARE: Conservative Model-Based Reward Learning for Offline Inverse Reinforcement Learning (6 citations)
• Learning Soft Constraints From Constrained Expert Demonstrations (3 citations)
• Benchmarking Constraint Inference in Inverse Reinforcement Learning (2 citations)
• Seeing Differently, Acting Similarly: Heterogeneously Observable Imitation Learning (2 citations)
• BC-IRL: Learning Generalizable Reward Functions from Demonstrations (1 citation)
• LS-IQ: Implicit Reward Regularization for Inverse Reinforcement Learning (1 citation)
• Visual Imitation Learning with Patch Rewards (1 citation)
• Adversarial Imitation Learning with Preferences
• Causal Imitation Learning via Inverse Reinforcement Learning
• Impossibly Good Experts and How to Follow Them
• Understanding and Adopting Rational Behavior by Bellman Score Estimation

RL from feedback · 4 papers

• Is Reinforcement Learning (Not) for Natural Language Processing: Benchmarks, Baselines, and Building Blocks for Natural Language Policy Optimization (53 citations)
• Reward Design with Language Models (21 citations)
• Preference Transformer: Modeling Human Preferences using Transformers for RL (4 citations)
• Causal Confusion and Reward Misidentification in Preference-Based Reward Learning (3 citations)