Model-free Posterior Sampling via Learning Rate Randomization
This addresses the problem of efficient exploration in reinforcement learning for researchers and practitioners, offering a novel method that is incremental in its algorithmic design.
The paper tackles regret minimization in episodic Markov Decision Processes by introducing Randomized Q-learning (RandQL), a model-free posterior sampling algorithm that achieves regret bounds of order $\widetilde{O}(\sqrt{H^{5}SAT})$ in tabular settings and $\widetilde{O}(H^{5/2} T^{(d_z+1)/(d_z+2)})$ in metric spaces, and empirically outperforms existing approaches.
In this paper, we introduce Randomized Q-learning (RandQL), a novel randomized model-free algorithm for regret minimization in episodic Markov Decision Processes (MDPs). To the best of our knowledge, RandQL is the first tractable model-free posterior sampling-based algorithm. We analyze the performance of RandQL in both tabular and non-tabular metric space settings. In tabular MDPs, RandQL achieves a regret bound of order $\widetilde{O}(\sqrt{H^{5}SAT})$, where $H$ is the planning horizon, $S$ is the number of states, $A$ is the number of actions, and $T$ is the number of episodes. For a metric state-action space, RandQL enjoys a regret bound of order $\widetilde{O}(H^{5/2} T^{(d_z+1)/(d_z+2)})$, where $d_z$ denotes the zooming dimension. Notably, RandQL achieves optimistic exploration without using bonuses, relying instead on a novel idea of learning rate randomization. Our empirical study shows that RandQL outperforms existing approaches on baseline exploration environments.