TarMAC: Targeted Multi-Agent Communication

Abhishek Das, Théophile Gervet, Joshua Romoff, Dhruv Batra, Devi Parikh, Mike Rabbat, Joelle Pineau

Abstract: We propose a targeted communication architecture for multi-agent reinforcement learning, where agents learn both <em>what</em> messages to send and <em>whom</em> to address them to while performing...