The $f$-Divergence Reinforcement Learning Framework
This provides a novel framework for reinforcement learning that addresses overestimation issues and improves efficiency, though it appears incremental as it builds on existing DRL methods.
The paper tackles the problem of sequential decision-making in deep reinforcement learning by introducing the $f$-Divergence Reinforcement Learning (FRL) framework, which simultaneously performs policy evaluation and improvement by minimizing $f$-divergence, and shows that agents trained with FRL match or surpass baseline algorithms on Atari 2600 games.
The framework of deep reinforcement learning (DRL) provides a powerful and widely applicable mathematical formalization for sequential decision-making. This paper present a novel DRL framework, termed \emph{$f$-Divergence Reinforcement Learning (FRL)}. In FRL, the policy evaluation and policy improvement phases are simultaneously performed by minimizing the $f$-divergence between the learning policy and sampling policy, which is distinct from conventional DRL algorithms that aim to maximize the expected cumulative rewards. We theoretically prove that minimizing such $f$-divergence can make the learning policy converge to the optimal policy. Besides, we convert the process of training agents in FRL framework to a saddle-point optimization problem with a specific $f$ function through Fenchel conjugate, which forms new methods for policy evaluation and policy improvement. Through mathematical proofs and empirical evaluation, we demonstrate that the FRL framework has two advantages: (1) policy evaluation and policy improvement processes are performed simultaneously and (2) the issues of overestimating value function are naturally alleviated. To evaluate the effectiveness of the FRL framework, we conduct experiments on Atari 2600 video games and show that agents trained in the FRL framework match or surpass the baseline DRL algorithms.