Credit-cognisant reinforcement learning for multi-agent cooperation

TL;DR

This paper introduces credit-cognisant rewards (CCRs) in multi-agent reinforcement learning to improve cooperation and performance in partially observable environments, demonstrated through a card game benchmark.

Contribution

The paper proposes CCRs to enhance credit assignment in MARL, outperforming traditional independent and recurrent Q-learning methods.

Findings

CCRs significantly improve MARL performance.

Enhanced credit assignment leads to better cooperation.

Results demonstrated on a simplified Hanabi game.

Abstract

Traditional multi-agent reinforcement learning (MARL) algorithms, such as independent Q-learning, struggle when presented with partially observable scenarios, and where agents are required to develop delicate action sequences. This is often the result of the reward for a good action only being available after other agents have taken theirs, and these actions are not credited accordingly. Recurrent neural networks have proven to be a viable solution strategy for solving these types of problems, resulting in significant performance increase when compared to other methods. In this paper, we explore a different approach and focus on the experiences used to update the action-value functions of each agent. We introduce the concept of credit-cognisant rewards (CCRs), which allows an agent to perceive the effect its actions had on the environment as well as on its co-agents. We show that by manipulating these experiences and constructing the reward contained within them to include the rewards received by all the agents within the same action sequence, we are able to improve significantly on the performance of independent deep Q-learning as well as deep recurrent Q-learning. We evaluate and test the performance of CCRs when applied to deep reinforcement learning techniques at the hands of a simplified version of the popular card game Hanabi.