An In-Depth Analysis of Discretization Methods for Communication Learning using Backpropagation with Multi-Agent Reinforcement Learning

TL;DR

This paper compares various discretization methods for communication in multi-agent reinforcement learning, introduces a novel method called ST-DRU, and demonstrates its superior performance across multiple environments.

Contribution

It provides a comprehensive comparison of discretization techniques and introduces ST-DRU, a new method that outperforms existing approaches in complex multi-agent settings.

Findings

ST-DRU achieves the best results across environments.

ST-DRU is the only method that does not fail in any environment.

The paper extends COMA-DIAL for complex environment testing.

Abstract

Communication is crucial in multi-agent reinforcement learning when agents are not able to observe the full state of the environment. The most common approach to allow learned communication between agents is the use of a differentiable communication channel that allows gradients to flow between agents as a form of feedback. However, this is challenging when we want to use discrete messages to reduce the message size, since gradients cannot flow through a discrete communication channel. Previous work proposed methods to deal with this problem. However, these methods are tested in different communication learning architectures and environments, making it hard to compare them. In this paper, we compare several state-of-the-art discretization methods as well as a novel approach. We do this comparison in the context of communication learning using gradients from other agents and perform tests on several environments. In addition, we present COMA-DIAL, a communication learning approach based on DIAL and COMA extended with learning rate scaling and adapted exploration. Using COMA-DIAL allows us to perform experiments on more complex environments. Our results show that the novel ST-DRU method, proposed in this paper, achieves the best results out of all discretization methods across the different environments. It achieves the best or close to the best performance in each of the experiments and is the only method that does not fail on any of the tested environments.