Low-Bandwidth Communication Emerges Naturally in Multi-Agent Learning Systems
This work provides a computational model for low-bandwidth emergent communication in multi-agent systems, which is relevant for researchers studying artificial intelligence and multi-agent learning.
This paper investigates emergent communication in multi-agent systems, specifically focusing on low-bandwidth communication inspired by natural cooperative behaviors. Through pursuit-evasion games, the authors demonstrate that multi-agent reinforcement learning algorithms can model this low-bandwidth communication.
In this work, we study emergent communication through the lens of cooperative multi-agent behavior in nature. Using insights from animal communication, we propose a spectrum from low-bandwidth (e.g. pheromone trails) to high-bandwidth (e.g. compositional language) communication that is based on the cognitive, perceptual, and behavioral capabilities of social agents. Through a series of experiments with pursuit-evasion games, we identify multi-agent reinforcement learning algorithms as a computational model for the low-bandwidth end of the communication spectrum.