Distributed Nash equilibrium-seeking Reinforcement Learning for N-player Games

点击翻译按钮获取中文摘要

Understanding and coordinating cooperation and competition between multiple intelligent decision-makers with shared resources is perhaps the most crucial challenge of our connected society. Game theory provides powerful tools to analyze such strategic decision-making among a set of agents. It is especially attractive for decision-making in autonomous systems and networks. Such systems are ubiquitous around us - they are spanning engineered, physical, biological and social systems. Many problems in these systems can be modeled as multi-player games. Examples range from congestion control in communication networks, ad-hoc (peer-to-peer) wireless networks, to demand-side management in smart grid/power networks, groups of autonomous robots engaged in search and rescue missions, and even interaction of users over social media. This project is focused on learning and reinforcement learning algorithms that find Nash equilibrium in multi-player (N-player) games, as well as providing the theory for such algorithms. The problem that the project aims to study is fundamental to multi-agent systems, where information between agents/players may be complete or incomplete. Multi-agent systems may be the next big area where reinforcement learning will have a big breakthrough. However, the theory in this domain has remained very little explored. This research project's goal is to advance the development of such theory and its potential applications in information and communication technologies (ICT). The project represents the partnership between the group of Professor Pavel, an internationally recognized researcher in game theory, and Huwaei, a world leader in information and communication technologies, with strong presence in Canada.