Extreme Q-Learning: MaxEnt RL without Entropy

Divyansh Garg
Donald Joseph Hejna III
Matthieu Geist
Stefano Ermon
International Conference on Representation Learning (ICLR)(2023)
Google Scholar


Modern Deep Reinforcement Learning (RL) algorithms require estimates of the maximal Q-value, which are difficult to compute in continuous domains with an infinite number of possible actions. In this work, we introduce a new update rule for online and offline RL which directly models the maximal value using Extreme Value Theory (EVT) inspired by Economics. By doing so, we avoid computing Q-values using out-of-distribution actions which is often a substantial source of error. Our key insight is to introduce an objective that directly estimates the optimal soft-value functions (LogSumExp) in the maximum entropy (MaxEnt) RL setting without needing to sample from a policy. Using EVT, we derive our \emph{Extreme Q-Learning} framework and consequently online and, for the first time, offline MaxEnt Q-learning algorithms, that do not explicitly require access to a policy or its entropy. Finally, our method obtains strong results in the Offline D4RL benchmark outperforming prior works by 10-20 points on some tasks while offering moderate improvements over SAC and TD3 on online DM Control tasks.

Research Areas