NeurIPS 2022 accepted papers on Reinforcement learning
232 of the 2,901 papers accepted at NeurIPS 2022 (8.0%) are on Reinforcement learning, compared with 8.6% at NeurIPS 2021. They fall into 6 topics.
See NeurIPS 2022 in the interactive map
Submitting to NeurIPS? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Visual and representation RL · 92 papers
- Pre-Trained Language Models for Interactive Decision-Making (53 citations)
- FinRL-Meta: Market Environments and Benchmarks for Data-Driven Financial Reinforcement Learning (48 citations)
- Towards Human-Level Bimanual Dexterous Manipulation with Reinforcement Learning (28 citations)
- Contrastive Learning as Goal-Conditioned Reinforcement Learning (24 citations)
- Mask-based Latent Reconstruction for Reinforcement Learning (21 citations)
- MaskPlace: Fast Chip Placement via Reinforced Visual Representation Learning (21 citations)
- On the Effectiveness of Fine-tuning Versus Meta-reinforcement Learning (19 citations)
- Discovered Policy Optimisation (16 citations)
- Improving Intrinsic Exploration with Language Abstractions (16 citations)
- Semantic Exploration from Language Abstractions and Pretrained Representations (16 citations)
- Reinforcement Learning with Neural Radiance Fields (15 citations)
- Deep Hierarchical Planning from Pixels (14 citations)
and 80 more in the interactive map.
Policy gradient and actor-critic · 57 papers
- Constrained Update Projection Approach to Safe Policy Optimization (19 citations)
- Model-based Safe Deep Reinforcement Learning via a Constrained Proximal Policy Optimization Algorithm (19 citations)
- Markovian Interference in Experiments (18 citations)
- Towards Safe Reinforcement Learning with a Safety Editor Policy (12 citations)
- TTOpt: A Maximum Volume Quantized Tensor Train-based Optimization and its Application to Reinforcement Learning (9 citations)
- Efficient Adversarial Training without Attacking: Worst-Case-Aware Robust Reinforcement Learning (8 citations)
- Enhancing Safe Exploration Using Safety State Augmentation (8 citations)
- When to Update Your Model: Constrained Model-based Reinforcement Learning (7 citations)
- Distinguishing Learning Rules with Brain Machine Interfaces (5 citations)
- Efficient Risk-Averse Reinforcement Learning (5 citations)
- Monte Carlo Augmented Actor-Critic for Sparse Reward Deep Reinforcement Learning from Suboptimal Demonstrations (5 citations)
- A Simple Decentralized Cross-Entropy Method (4 citations)
and 45 more in the interactive map.
Offline RL · 30 papers
- Mildly Conservative Q-Learning for Offline Reinforcement Learning (26 citations)
- NeoRL: A Near Real-World Benchmark for Offline Reinforcement Learning (24 citations)
- RAMBO-RL: Robust Adversarial Model-Based Offline Reinforcement Learning (18 citations)
- Robust Reinforcement Learning using Offline Data (15 citations)
- Bootstrapped Transformer for Offline Reinforcement Learning (13 citations)
- Double Check Your State Before Trusting It: Confidence-Aware Bidirectional Offline Model-Based Imagination (11 citations)
- Supported Policy Optimization for Offline Reinforcement Learning (11 citations)
- A Policy-Guided Imitation Approach for Offline Reinforcement Learning (10 citations)
- Exploit Reward Shifting in Value-Based Deep-RL: Optimistic Curiosity-Based Exploration and Conservative Exploitation via Linear Reward Shaping (10 citations)
- RORL: Robust Offline Reinforcement Learning via Conservative Smoothing (10 citations)
- You Can’t Count on Luck: Why Decision Transformers and RvS Fail in Stochastic Environments (8 citations)
- When to Trust Your Simulator: Dynamics-Aware Hybrid Offline-and-Online Reinforcement Learning (7 citations)
and 18 more in the interactive map.
MDPs and planning · 27 papers
- Learning to Branch with Tree MDPs (17 citations)
- Unpacking Reward Shaping: Understanding the Benefits of Reward Engineering on Sample Complexity (12 citations)
- DOPE: Doubly Optimistic and Pessimistic Exploration for Safe Reinforcement Learning (9 citations)
- Provable Defense against Backdoor Policies in Reinforcement Learning (7 citations)
- Robust Anytime Learning of Markov Decision Processes (4 citations)
- The Impact of Task Underspecification in Evaluating Deep Reinforcement Learning (4 citations)
- Instance-Dependent Near-Optimal Policy Identification in Linear MDPs via Online Experiment Design (3 citations)
- Learning in Observable POMDPs, without Computationally Intractable Oracles (3 citations)
- Model-based RL with Optimistic Posterior Sampling: Structural Conditions and Sample Complexity (3 citations)
- On Gap-dependent Bounds for Offline Reinforcement Learning (3 citations)
- Provably Efficient Reinforcement Learning in Partially Observable Dynamical Systems (3 citations)
- A Near-Optimal Primal-Dual Method for Off-Policy Learning in CMDP (2 citations)
and 15 more in the interactive map.
Imitation learning · 21 papers
- Active Exploration for Inverse Reinforcement Learning (6 citations)
- Improving Policy Learning via Language Dynamics Distillation (5 citations)
- Chain of Thought Imitation with Procedure Cloning (4 citations)
- Maximum-Likelihood Inverse Reinforcement Learning with Finite-Time Guarantees (4 citations)
- Robust Imitation of a Few Demonstrations with a Backwards Model (4 citations)
- TarGF: Learning Target Gradient Field to Rearrange Objects without Explicit Goal Specification (4 citations)
- TaSIL: Taylor Series Imitation Learning (4 citations)
- Imitating Past Successes can be Very Suboptimal (3 citations)
- Sequence Model Imitation Learning with Unobserved Contexts (3 citations)
- Dynamic Inverse Reinforcement Learning for Characterizing Animal Behavior (2 citations)
- Planning for Sample Efficient Imitation Learning (2 citations)
- CEIP: Combining Explicit and Implicit Priors for Reinforcement Learning with Demonstrations (1 citation)
and 9 more in the interactive map.
RL from feedback · 5 papers
- Non-Markovian Reward Modelling from Trajectory Labels via Interpretable Multiple Instance Learning (4 citations)
- Teacher Forcing Recovers Reward Functions for Text Generation (4 citations)
- Meta-Reward-Net: Implicitly Differentiable Reward Learning for Preference-based Reinforcement Learning (1 citation)
- Grounded Reinforcement Learning: Learning to Win the Game under Human Commands
- On Reinforcement Learning and Distribution Matching for Fine-Tuning Language Models with no Catastrophic Forgetting
