Online Statistical Inference of Constant Sample-averaged Q-Learning
This work addresses the need for reliable uncertainty quantification in reinforcement learning, particularly for practitioners who require confidence intervals for Q-values in decision-making tasks.
The paper proposes a framework for online statistical inference of sample-averaged Q-learning, adapting the functional central limit theorem to construct confidence intervals for Q-values via random scaling. Experiments on a grid world and a dynamic resource-matching problem show that the modified approach provides valid coverage rates and narrower confidence intervals compared to traditional Q-learning.
Reinforcement learning algorithms have been widely used for decision-making tasks in various domains. However, the performance of these algorithms can be impacted by high variance and instability, particularly in environments with noise or sparse rewards. In this paper, we propose a framework to perform statistical online inference for a sample-averaged Q-learning approach. We adapt the functional central limit theorem (FCLT) for the modified algorithm under some general conditions and then construct confidence intervals for the Q-values via random scaling. We conduct experiments to perform inference on both the modified approach and its traditional counterpart, Q-learning using random scaling and report their coverage rates and confidence interval widths on two problems: a grid world problem as a simple toy example and a dynamic resource-matching problem as a real-world example for comparison between the two solution approaches.