overview 

In recent years, we have witnessed a great success of AI in many applications, including image classification, recommendation systems, etc. Since machine learning models are deployed in the real world, these models will interact with and impact each other, turning their decision-making into a multi-agent problem. Therefore, multi-agent learning in a complex world is a fundamental problem for the next generation of AI to empower various multi-agent tasks, among which, cooperative tasks are of the first and foremost interest to practitioners.  In this tutorial, we will give a thorough review of fundamentals, progress, and challenges for cooperative multi-agent learning, including 1) the basics of reinforcement learning, multi-agent sequential decision making, 2) the research problems including scalability, decentralization, coordination, and a review of progresses, 3) cooperative learning for self-interested agents, and 4) directions for future works. 

The prior knowledge required is the basics of reinforcement learning and moderate experience in machine learning, which enables the audience to be comfortable in following the discussion of the related works and progress.

OUTLINE

The tutorial is split into two main sessions:

• Welcome (2min).

• Background of multi-agent learning (~10 min), presented by Yali Du.

• Team-based cooperative learning (~40 min), presented by Yali Du.

• Self-interested cooperative learning (~50 min), presented by Joel Z. Leibo.

• Conclusions (~5 min).

SLIDES

To be added

PRESENTERS