2020-10-14 | Xiang Yan：Internet Market Design: from Algorithmic Game Theory to Machine Learning
Recent years witnessed a burst of online business such as network resource sharing, data selling, and ads auctions. These online businesses naturally fall into a multi-agent system where agents make decisions and interact with each other to achieve their own objectives, while at the same time, possibly manipulate the mechanisms designed by market makers. This talk begins with a traditional P2P resource sharing market, analyzing the potential cheating behaviors by strategic agents and showing the robustness of the market equilibrium under a proportional response mechanism. Then the incentive analysis is extended to special data markets, with machine learners being the market designers, and algorithms with incentive guarantee are provided. Finally, in the competition between multiple Internet market makers, machine learning tools are successfully adopted to deal with the latent information in such a game scenario.
Xiang Yan got his BSc from Zhiyuan College in Shanghai Jiao Tong University in 2015. He is currently a PhD candidate of Department of Computer Science in Shanghai Jiao Tong University, and a visiting scholar of School of Engineering and Applied Sciences in Harvard University.
Yan's current research focuses on algorithmic game theory and machine learning, with their applications to Internet Economics.