Provably Efficient Exploration in Policy Optimization
This provides a provably efficient algorithm for policy optimization with exploration, addressing a gap in theory for RL practitioners.
The paper tackles the lack of theoretical understanding for policy-based reinforcement learning by proposing OPPO, an optimistic variant of Proximal Policy Optimization, which achieves a regret bound of $ ilde{O}(\sqrt{d^2 H^3 T})$ in episodic Markov decision processes with linear function approximation.
While policy-based reinforcement learning (RL) achieves tremendous successes in practice, it is significantly less understood in theory, especially compared with value-based RL. In particular, it remains elusive how to design a provably efficient policy optimization algorithm that incorporates exploration. To bridge such a gap, this paper proposes an Optimistic variant of the Proximal Policy Optimization algorithm (OPPO), which follows an ``optimistic version'' of the policy gradient direction. This paper proves that, in the problem of episodic Markov decision process with linear function approximation, unknown transition, and adversarial reward with full-information feedback, OPPO achieves $\tilde{O}(\sqrt{d^2 H^3 T} )$ regret. Here $d$ is the feature dimension, $H$ is the episode horizon, and $T$ is the total number of steps. To the best of our knowledge, OPPO is the first provably efficient policy optimization algorithm that explores.