Joint Policy Search for Multi-agent Collaboration with Imperfect Information

To learn good joint policies for multi-agent collaboration with imperfect information remains a fundamental challenge.