Advantage-Weighted Regression: Simple and Scalable Off-Policy Reinforcement Learning

Abstract

In this paper, we aim to develop a simple and scalable reinforcement learning algorithm that uses standard supervised learning methods as subroutines. Our goal is an algorithm that utilizes only simple and convergent maximum likelihood loss functions, while also being able to leverage off-policy data. Our proposed approach, which we refer to as advantage-weighted regression (AWR), consists of two standard supervised learning steps: one to regress onto target values for a value function, and another to regress onto weighted target actions for the policy. The method is simple and general, can accommodate continuous and discrete actions, and can be implemented in just a few lines of code on top of standard supervised learning methods. We provide a theoretical motivation for AWR and analyze its properties when incorporating off-policy data from experience replay. We evaluate AWR on a suite of standard OpenAI Gym benchmark tasks, and show that it achieves competitive performance compared to a number of well-established state-of-the-art RL algorithms. AWR is also able to acquire more effective policies than most off-policy algorithms when learning from purely static datasets with no additional environmental interactions. Furthermore, we demonstrate our algorithm on challenging continuous control tasks with highly complex simulated characters.

References in corpus (1)

• Emergence of Locomotion Behaviours in Rich Environments

Cited by in corpus (29)

• How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learned
• Conservative Q-Learning for Offline Reinforcement Learning
• Decision Transformer: Reinforcement Learning via Sequence Modeling
• Offline Reinforcement Learning with Implicit Q-Learning
• What Matters In On-Policy Reinforcement Learning? A Large-Scale Empirical Study
• Phasic Policy Gradient
• COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
• Learning Deformable Object Manipulation from Expert Demonstrations
• Rewriting History with Inverse RL: Hindsight Inference for Policy Improvement
• Uncertainty Weighted Actor-Critic for Offline Reinforcement Learning
• Believe What You See: Implicit Constraint Approach for Offline Multi-Agent Reinforcement Learning
• Reward-Conditioned Policies
• Offline Reinforcement Learning from Images with Latent Space Models
• Offline Learning from Demonstrations and Unlabeled Experience
• PLAS: Latent Action Space for Offline Reinforcement Learning
• Human-in-the-Loop Imitation Learning using Remote Teleoperation
• Offline Reinforcement Learning with Value-based Episodic Memory
• Learning and Planning in Complex Action Spaces
• Offline Reinforcement Learning with Soft Behavior Regularization
• Behavioral Priors and Dynamics Models: Improving Performance and Domain Transfer in Offline RL
• Skill Preferences: Learning to Extract and Execute Robotic Skills from Human Feedback
• Semi-supervised reward learning for offline reinforcement learning
• Instabilities of Offline RL with Pre-Trained Neural Representation
• A Workflow for Offline Model-Free Robotic Reinforcement Learning
• Safe Driving via Expert Guided Policy Optimization
• Conservative Data Sharing for Multi-Task Offline Reinforcement Learning
• On Covariate Shift of Latent Confounders in Imitation and Reinforcement Learning
• You Only Evaluate Once: a Simple Baseline Algorithm for Offline RL
• Quantile Filtered Imitation Learning