Learning in Cooperative Multiagent Systems Using Cognitive and Machine Models

TL;DR

This paper introduces Multi-Agent IBL models that combine cognitive theories with deep reinforcement learning to improve coordination and learning speed in stochastic multi-agent environments, outperforming existing MADRL approaches.

Contribution

The paper proposes three variants of MAIBL models integrating cognitive mechanisms with MADRL to enhance learning and coordination in stochastic multi-agent tasks.

Findings

MAIBL models achieve faster learning than MADRL.

MAIBL models demonstrate better coordination in CMOTP.

Integration of cognitive insights benefits MADRL performance.

Abstract

Developing effective Multi-Agent Systems (MAS) is critical for many applications requiring collaboration and coordination with humans. Despite the rapid advance of Multi-Agent Deep Reinforcement Learning (MADRL) in cooperative MAS, one major challenge is the simultaneous learning and interaction of independent agents in dynamic environments in the presence of stochastic rewards. State-of-the-art MADRL models struggle to perform well in Coordinated Multi-agent Object Transportation Problems (CMOTPs), wherein agents must coordinate with each other and learn from stochastic rewards. In contrast, humans often learn rapidly to adapt to nonstationary environments that require coordination among people. In this paper, motivated by the demonstrated ability of cognitive models based on Instance-Based Learning Theory (IBLT) to capture human decisions in many dynamic decision making tasks, we propose three variants of Multi-Agent IBL models (MAIBL). The idea of these MAIBL algorithms is to combine the cognitive mechanisms of IBLT and the techniques of MADRL models to deal with coordination MAS in stochastic environments from the perspective of independent learners. We demonstrate that the MAIBL models exhibit faster learning and achieve better coordination in a dynamic CMOTP task with various settings of stochastic rewards compared to current MADRL models. We discuss the benefits of integrating cognitive insights into MADRL models.