ICLR 2022 accepted papers on Reinforcement learning
125 of the 1,095 papers accepted at ICLR 2022 (11.4%) are on Reinforcement learning, compared with 10.9% at ICLR 2021. They fall into 6 topics.
See ICLR 2022 in the interactive map
Submitting to ICLR? 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 · 65 papers
- Learning Vision-Guided Quadrupedal Locomotion End-to-End with Cross-Modal Transformers (25 citations)
- The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models (21 citations)
- Accelerated Policy Learning with Parallel Differentiable Simulation (15 citations)
- DiffSkill: Skill Abstraction from Differentiable Physics for Deformable Object Manipulations with Tools (12 citations)
- Skill-based Meta-Reinforcement Learning (11 citations)
- Understanding and Preventing Capacity Loss in Reinforcement Learning (10 citations)
- CoBERL: Contrastive BERT for Reinforcement Learning (8 citations)
- Lipschitz-constrained Unsupervised Skill Discovery (8 citations)
- Active Hierarchical Exploration with Stable Subgoal Representation Learning (7 citations)
- Modular Lifelong Reinforcement Learning via Neural Composition (7 citations)
- AdaRL: What, Where, and How to Adapt in Transfer Reinforcement Learning (6 citations)
- Cross-Trajectory Representation Learning for Zero-Shot Generalization in RL (6 citations)
and 53 more in the interactive map.
Policy gradient and actor-critic · 21 papers
- Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learning (63 citations)
- Maximum Entropy RL (Provably) Solves Some Robust RL Problems (28 citations)
- Mirror Descent Policy Optimization (23 citations)
- Dropout Q-Functions for Doubly Efficient Reinforcement Learning (19 citations)
- Sample Efficient Deep Reinforcement Learning via Uncertainty Estimation (15 citations)
- Value Gradient weighted Model-Based Reinforcement Learning (5 citations)
- $\mathrm{SO}(2)$-Equivariant Reinforcement Learning (4 citations)
- CROP: Certifying Robust Policies for Reinforcement Learning through Functional Smoothing (4 citations)
- Actor-critic is implicitly biased towards high entropy optimal policies (1 citation)
- Orchestrated Value Mapping for Reinforcement Learning (1 citation)
- Overcoming The Spectral Bias of Neural Value Approximation (1 citation)
- Is High Variance Unavoidable in RL? A Case Study in Continuous Control
and 9 more in the interactive map.
Offline RL · 18 papers
- Offline Reinforcement Learning with Implicit Q-Learning (130 citations)
- Reinforcement Learning with Sparse Rewards using Guidance from Offline Demonstration (27 citations)
- Pessimistic Bootstrapping for Uncertainty-Driven Offline Reinforcement Learning (24 citations)
- Rethinking Goal-Conditioned Supervised Learning and Its Connection to Offline RL (13 citations)
- Offline Reinforcement Learning with Value-based Episodic Memory (10 citations)
- DARA: Dynamics-Aware Reward Augmentation in Offline Reinforcement Learning (8 citations)
- DR3: Value-Based Deep Reinforcement Learning Requires Explicit Regularization (6 citations)
- Pessimistic Model-based Offline Reinforcement Learning under Partial Coverage (6 citations)
- Revisiting Design Choices in Offline Model Based Reinforcement Learning (6 citations)
- Generalized Decision Transformer for Offline Hindsight Information Matching (5 citations)
- COptiDICE: Offline Constrained Reinforcement Learning via Stationary Distribution Correction Estimation (3 citations)
- Model-Based Offline Meta-Reinforcement Learning with Regularization (3 citations)
and 6 more in the interactive map.
MDPs and planning · 11 papers
- Understanding Domain Randomization for Sim-to-real Transfer (33 citations)
- Constrained Policy Optimization via Bayesian World Models (10 citations)
- On the Convergence of the Monte Carlo Exploring Starts Algorithm for Reinforcement Learning (5 citations)
- Representation Learning for Online and Offline RL in Low-rank MDPs (5 citations)
- Towards Deployment-Efficient Reinforcement Learning: Lower Bound and Optimality (4 citations)
- An Experimental Design Perspective on Model-Based Reinforcement Learning (3 citations)
- Near-optimal Offline Reinforcement Learning with Linear Representation: Leveraging Variance Information with Pessimism (3 citations)
- The Geometry of Memoryless Stochastic Policy Optimization in Infinite-Horizon POMDPs (3 citations)
- Near-Optimal Reward-Free Exploration for Linear Mixture MDPs with Plug-in Solver (2 citations)
- Learning Value Functions from Undirected State-only Experience (1 citation)
- Online Target Q-learning with Reverse Experience Replay: Efficiently finding the Optimal Policy for Linear MDPs
Imitation learning · 8 papers
- The Boltzmann Policy Distribution: Accounting for Systematic Suboptimality in Human Models (7 citations)
- Hierarchical Few-Shot Imitation with Skill Transition Models (6 citations)
- TRAIL: Near-Optimal Imitation Learning with Suboptimal Data (5 citations)
- On Covariate Shift of Latent Confounders in Imitation and Reinforcement Learning (3 citations)
- Imitation Learning by Reinforcement Learning
- Imitation Learning from Observations under Transition Model Disparity
- CrowdPlay: Crowdsourcing Human Demonstrations for Offline Learning
- DemoDICE: Offline Imitation Learning with Supplementary Imperfect Demonstrations
