MEET: A Monte Carlo Exploration-Exploitation Trade-off for Buffer Sampling

TL;DR

This paper introduces MEET, a novel buffer sampling method for reinforcement learning that uses Q-value uncertainty to balance exploration and exploitation, leading to faster and better policy learning.

Contribution

It proposes a new sampling strategy leveraging Q-value uncertainty to adapt exploration-exploitation trade-offs in experience replay buffers.

Findings

Outperforms state-of-the-art sampling strategies by 26% on average.

Demonstrates stable and improved convergence in classical control environments.

Enhances learning efficiency by focusing on significant transitions.

Abstract

Data selection is essential for any data-based optimization technique, such as Reinforcement Learning. State-of-the-art sampling strategies for the experience replay buffer improve the performance of the Reinforcement Learning agent. However, they do not incorporate uncertainty in the Q-Value estimation. Consequently, they cannot adapt the sampling strategies, including exploration and exploitation of transitions, to the complexity of the task. To address this, this paper proposes a new sampling strategy that leverages the exploration-exploitation trade-off. This is enabled by the uncertainty estimation of the Q-Value function, which guides the sampling to explore more significant transitions and, thus, learn a more efficient policy. Experiments on classical control environments demonstrate stable results across various environments. They show that the proposed method outperforms state-of-the-art sampling strategies for dense rewards w.r.t. convergence and peak performance by 26% on average.