Temporal Difference Models: Model-Free Deep RL for Model-Based Control

TL;DR

Temporal Difference Models (TDMs) integrate model-free and model-based reinforcement learning to improve sample efficiency and asymptotic performance in continuous control tasks by leveraging rich transition information.

Contribution

Introduction of TDMs, goal-conditioned value functions trained with model-free methods that enable efficient model-based control, surpassing existing approaches.

Findings

TDMs significantly outperform state-of-the-art methods in continuous control tasks.

TDMs achieve higher sample efficiency than traditional model-free RL.

TDMs combine advantages of model-free and model-based RL for better performance.

Abstract

Model-free reinforcement learning (RL) is a powerful, general tool for learning complex behaviors. However, its sample efficiency is often impractically large for solving challenging real-world problems, even with off-policy algorithms such as Q-learning. A limiting factor in classic model-free RL is that the learning signal consists only of scalar rewards, ignoring much of the rich information contained in state transition tuples. Model-based RL uses this information, by training a predictive model, but often does not achieve the same asymptotic performance as model-free RL due to model bias. We introduce temporal difference models (TDMs), a family of goal-conditioned value functions that can be trained with model-free learning and used for model-based control. TDMs combine the benefits of model-free and model-based RL: they leverage the rich information in state transitions to learn very efficiently, while still attaining asymptotic performance that exceeds that of direct model-based RL methods. Our experimental results show that, on a range of continuous control tasks, TDMs provide a substantial improvement in efficiency compared to state-of-the-art model-based and model-free methods.