Towards Heterogeneous Multi-Agent Reinforcement Learning with Graph Neural Networks

Abstract

This work proposes a neural network architecture that learns policies for multiple agent classes in a heterogeneous multi-agent reinforcement setting. The proposed network uses directed labeled graph representations for states, encodes feature vectors of different sizes for different entity classes, uses relational graph convolution layers to model different communication channels between entity types and learns distinct policies for different agent classes, sharing parameters wherever possible. Results have shown that specializing the communication channels between entity classes is a promising step to achieve higher performance in environments composed of heterogeneous entities.

arXiv commentPublished and presented at the XVI Encontro Nacional de Inteligência Artificial e Computacional (ENIAC). Fixed notation

References in corpus (5)

• Semi-Supervised Classification with Graph Convolutional Networks
• Neural Message Passing for Quantum Chemistry
• Convolutional Networks on Graphs for Learning Molecular Fingerprints
• Learning Attentional Communication for Multi-Agent Cooperation
• Multiagent Bidirectionally-Coordinated Nets: Emergence of Human-level Coordination in Learning to Play StarCraft Combat Games

Cited by in corpus (1)

• Heterogeneous Graph Attention Networks for Learning Diverse Communication