Learning to Ground Decentralized Multi-Agent Communication with Contrastive Learning
read the original abstract
For communication to happen successfully, a common language is required between agents to understand information communicated by one another. Inducing the emergence of a common language has been a difficult challenge to multi-agent learning systems. In this work, we introduce an alternative perspective to the communicative messages sent between agents, considering them as different incomplete views of the environment state. Based on this perspective, we propose a simple approach to induce the emergence of a common language by maximizing the mutual information between messages of a given trajectory in a self-supervised manner. By evaluating our method in communication-essential environments, we empirically show how our method leads to better learning performance and speed, and learns a more consistent common language than existing methods, without introducing additional learning parameters.
This paper has not been read by Pith yet.
Forward citations
Cited by 1 Pith paper
-
CCKS: Consensus-based Communication and Knowledge Sharing
CCKS adds consensus constraints built by contrastive learning on local observations to action-advising in DTDE MARL, yielding faster learning and higher performance on football and StarCraft benchmarks.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.