Adaptive Ensemble Q-learning: Minimizing Estimation Bias via Error Feedback

TL;DR

This paper introduces AdaEQ, an adaptive ensemble Q-learning method that dynamically adjusts the ensemble size based on error feedback to minimize estimation bias and improve learning performance in continuous control tasks.

Contribution

The paper proposes a novel adaptive ensemble Q-learning algorithm that uses theoretical bounds and error feedback to optimize ensemble size during training.

Findings

AdaEQ reduces estimation bias effectively.

AdaEQ outperforms existing methods on MuJoCo benchmarks.

Dynamic ensemble sizing improves learning stability.

Abstract

The ensemble method is a promising way to mitigate the overestimation issue in Q-learning, where multiple function approximators are used to estimate the action values. It is known that the estimation bias hinges heavily on the ensemble size (i.e., the number of Q-function approximators used in the target), and that determining the `right' ensemble size is highly nontrivial, because of the time-varying nature of the function approximation errors during the learning process. To tackle this challenge, we first derive an upper bound and a lower bound on the estimation bias, based on which the ensemble size is adapted to drive the bias to be nearly zero, thereby coping with the impact of the time-varying approximation errors accordingly. Motivated by the theoretic findings, we advocate that the ensemble method can be combined with Model Identification Adaptive Control (MIAC) for effective ensemble size adaptation. Specifically, we devise Adaptive Ensemble Q-learning (AdaEQ), a generalized ensemble method with two key steps: (a) approximation error characterization which serves as the feedback for flexibly controlling the ensemble size, and (b) ensemble size adaptation tailored towards minimizing the estimation bias. Extensive experiments are carried out to show that AdaEQ can improve the learning performance than the existing methods for the MuJoCo benchmark.