Mixture of Experts in a Mixture of RL settings

TL;DR

This paper investigates how Mixture of Experts (MoEs) can improve Deep Reinforcement Learning (DRL) in highly non-stationary, multi-task environments, providing new insights into their mechanisms and optimal integration.

Contribution

It offers a detailed analysis of MoEs' ability to handle non-stationarity in DRL, especially in multi-task settings, and explores how different MoE components influence learning.

Findings

MoEs enhance DRL performance in non-stationary, multi-task environments.

Multi-task experiments reveal the mechanisms behind MoEs' benefits.

Optimal MoE component integration improves actor-critic DRL networks.

Abstract

Mixtures of Experts (MoEs) have gained prominence in (self-)supervised learning due to their enhanced inference efficiency, adaptability to distributed training, and modularity. Previous research has illustrated that MoEs can significantly boost Deep Reinforcement Learning (DRL) performance by expanding the network's parameter count while reducing dormant neurons, thereby enhancing the model's learning capacity and ability to deal with non-stationarity. In this work, we shed more light on MoEs' ability to deal with non-stationarity and investigate MoEs in DRL settings with "amplified" non-stationarity via multi-task training, providing further evidence that MoEs improve learning capacity. In contrast to previous work, our multi-task results allow us to better understand the underlying causes for the beneficial effect of MoE in DRL training, the impact of the various MoE components, and insights into how best to incorporate them in actor-critic-based DRL networks. Finally, we also confirm results from previous work.