AI-driven mechanism design

发布时间:2019-07-05浏览次数:177

Speaker:   Dr. Weiran Shen

Time:       14:00-15:00, July 17

Location:  SIST 1A 301

Host:       Prof. Dengji Zhao

Abstract:

Due to its huge success in the industry, mechanism design has been one of the central research topics at the interface of economics and computer science. However, despite decades of efforts, we are still faced with challenges in both theory and practice. In theory, how to deal with huge design space and buyer strategy space still remains largely unknown, except for some specific cases. In practice, the most widely applied auction format (the generalized second price auction) is neither truthful nor optimal. And many theoretical results are built upon unrealistic assumptions that do not hold in real applications.

We propose the AI-driven mechanism design framework, aiming to provide an alternative way of dealing with these problems. By combining AI techniques with mechanism design theory, we are able to solve problems that cannot be solved using tools from either domain alone.

Bio:

Weiran Shen has just obtained his Ph.D. degree from IIIS, Tsinghua University, and will move to CMU for postdoc this summer. He is interested in doing research in the interdisciplinary area that combines computer science and economics, e.g., game theory, mechanism design, machine learning. During his Ph.D., he published 12 papers in CS conferences (AAAI, IJCAI, ICML, AAMAS, WWW). Mechanisms designed by him has already been adopted by Baidu, increased Baidu's revenue by over 5% and was highlighted in Baidu's 2018 Q1 Financial Report. 

SIST-Seminar 18190