Stochastic Q-learning for Large Discrete Action Spaces

TL;DR

This paper introduces stochastic value-based reinforcement learning methods that efficiently handle large discrete action spaces by considering only a small, stochastic subset of actions per iteration, reducing computational costs while maintaining high performance.

Contribution

The paper proposes novel stochastic Q-learning algorithms that consider a subset of actions, with proven convergence and superior empirical performance over traditional methods.

Findings

Outperforms baseline methods in diverse control environments

Achieves near-optimal returns with reduced computation time

Converges theoretically under certain conditions

Abstract

In complex environments with large discrete action spaces, effective decision-making is critical in reinforcement learning (RL). Despite the widespread use of value-based RL approaches like Q-learning, they come with a computational burden, necessitating the maximization of a value function over all actions in each iteration. This burden becomes particularly challenging when addressing large-scale problems and using deep neural networks as function approximators. In this paper, we present stochastic value-based RL approaches which, in each iteration, as opposed to optimizing over the entire set of $n$ actions, only consider a variable stochastic set of a sublinear number of actions, possibly as small as $\mathcal{O}(\log(n))$. The presented stochastic value-based RL methods include, among others, Stochastic Q-learning, StochDQN, and StochDDQN, all of which integrate this stochastic approach for both value-function updates and action selection. The theoretical convergence of Stochastic Q-learning is established, while an analysis of stochastic maximization is provided. Moreover, through empirical validation, we illustrate that the various proposed approaches outperform the baseline methods across diverse environments, including different control problems, achieving near-optimal average returns in significantly reduced time.