Non-Stationary Policy Learning for Multi-Timescale Multi-Agent Reinforcement Learning

TL;DR

This paper introduces a simple framework for learning non-stationary, multi-timescale policies in multi-agent reinforcement learning using periodic time encoding and phase-functioned neural networks, validated on gridworld and energy management tasks.

Contribution

It proposes a novel approach leveraging periodic encoding and phase-functioned neural networks to effectively learn non-stationary policies in multi-timescale MARL.

Findings

Successfully learned policies in gridworld environment.

Effective energy management policy in building environment.

Demonstrated theoretical learnability of non-stationary policies.

Abstract

In multi-timescale multi-agent reinforcement learning (MARL), agents interact across different timescales. In general, policies for time-dependent behaviors, such as those induced by multiple timescales, are non-stationary. Learning non-stationary policies is challenging and typically requires sophisticated or inefficient algorithms. Motivated by the prevalence of this control problem in real-world complex systems, we introduce a simple framework for learning non-stationary policies for multi-timescale MARL. Our approach uses available information about agent timescales to define a periodic time encoding. In detail, we theoretically demonstrate that the effects of non-stationarity introduced by multiple timescales can be learned by a periodic multi-agent policy. To learn such policies, we propose a policy gradient algorithm that parameterizes the actor and critic with phase-functioned neural networks, which provide an inductive bias for periodicity. The framework's ability to effectively learn multi-timescale policies is validated on a gridworld and building energy management environment.