Inference-Based Deterministic Messaging For Multi-Agent Communication

TL;DR

This paper introduces a deterministic messaging approach for multi-agent communication that improves convergence to optimal policies in decentralized settings, demonstrated through matrix signaling games and a cooperative gridworld environment.

Contribution

It proposes a deterministic sender policy modification that enhances convergence to optimal communication strategies in multi-agent systems.

Findings

Agents converge to optimal policies with the proposed method.

Deterministic messaging improves learning efficiency in signaling games.

Method extends to complex partially observable environments.

Abstract

Communication is essential for coordination among humans and animals. Therefore, with the introduction of intelligent agents into the world, agent-to-agent and agent-to-human communication becomes necessary. In this paper, we first study learning in matrix-based signaling games to empirically show that decentralized methods can converge to a suboptimal policy. We then propose a modification to the messaging policy, in which the sender deterministically chooses the best message that helps the receiver to infer the sender's observation. Using this modification, we see, empirically, that the agents converge to the optimal policy in nearly all the runs. We then apply this method to a partially observable gridworld environment which requires cooperation between two agents and show that, with appropriate approximation methods, the proposed sender modification can enhance existing decentralized training methods for more complex domains as well.