Towards Decentralized Reinforcement Learning Architectures for Social Dilemmas

Published in Autonomous Agents and Multi-Agent Systems, 2019

Multi-agent reinforcement learning has received significant interest in recent years notably due to the advancements made in deep reinforcement learning which have allowed for the developments of new architectures and learning algorithms. In this extended abstract we present our initial efforts towards the development of decentralized architectures for multi-agent systems in order to understand and model societies. More specifically, using social dilemmas as the training ground, we present a novel learning architecture, Learning through Probing (LTP), where agents utilize a probing mechanism to incorporate how their opponent’s behavior changes when an agent takes an action.

Authors: Nicolas Anastassacos, Mirco Musolesi

Link to arxiv here