Scalable Multi-agent Covering Option Discovery based on Kronecker Graphs
Published:
Codebase for my papers: Multi-agent Covering Option Discovery through Kronecker Product of Factor Graphs && Scalable Multi-agent Covering Option Discovery based on Kronecker Graphs
Language: Python
The following parts are included:
- Multi-agent maze tasks built with Mujoco as the benchmark for the continuous case.
- Multi-agent grid-maze tasks as the benchmark for the tabular case.
- Implementations of the multi-agent option discovery algorithms for tabular and deep MARL proposed in our paper.
- Implementations of tablar MARL algorithms as baselines: Independent Q-Learning, Distributed Q-Learning, Centralized Q-Learning.
- Implementations of Deep MARL algorithms as baselines: MAPPO, MAA2C, MADDPG, QMIX, Weighted-QMIX, MAVEN, COMA, MSAC.