Conservative Optimistic Policy Optimization via Multiple Importance Sampling

TL;DR

This paper introduces a conservative, model-free reinforcement learning algorithm with bounded regret, ensuring safer policy updates and improved reliability in real-world applications.

Contribution

It proposes a novel online policy optimization method that guarantees conservative exploration with regret bounds in both discrete and continuous settings.

Findings

Regret bounded by  ( ( ilde{O}(dT)) in various spaces

Algorithm ensures safer policy updates during reinforcement learning

Applicable to both discrete and continuous parameter spaces

Abstract

Reinforcement Learning (RL) has been able to solve hard problems such as playing Atari games or solving the game of Go, with a unified approach. Yet modern deep RL approaches are still not widely used in real-world applications. One reason could be the lack of guarantees on the performance of the intermediate executed policies, compared to an existing (already working) baseline policy. In this paper, we propose an online model-free algorithm that solves conservative exploration in the policy optimization problem. We show that the regret of the proposed approach is bounded by $\tilde{\mathcal{O}}(\sqrt{T})$ for both discrete and continuous parameter spaces.