Doctoral Thesis Oral Defense - Brian Hu Zhang

— 4:00pm

Location:
In Person and Virtual - ET - Reddy Conference Room, Gates Hillman 4405 and Zoom

Speaker:
BRIAN HU ZHANG , Ph.D. Candidate, Computer Science Department, Carnegie Mellon University
https://brianhzhang.github.io/

New Solution Concepts and Algorithms for Equilibrium Computation and Learning in Extensive-Form Games and Beyond

Computational game theory has led to significant breakthroughs in AI dating back to the start of AI as a discipline. For example, it has been instrumental in enabling superhuman AI from recreational games such as two-player zero-sum games chess, go, and heads-up poker to multiplayer games such as six-player poker and Hanabi, and even in games involving human language such as Diplomacy. It has also empowered a growing range of non-recreational applications, such as trading, machine learning robustness and safety, negotiation, conflict resolution, mechanism (e.g., auction) design, information design, security, political campaigning, and self-driving cars. 

This thesis pushes the boundary on computational game theory, especially in imperfect-information sequential (extensive-form) games, which are most prevalent in practical applications both in zero-sum games and beyond. We will present new theoretical concepts and frameworks, state-of-the-art and often provably optimal algorithms for computing and learning equilibria, and new ways to apply such algorithms to real-world problems, including problems in economics such as mechanism and information design. We will also draw connections to the broader literature on optimization, yielding new and more efficient algorithms for solving variational inequalities.

Thesis Committee

Tuomas Sandholm (Chair)
Vincent Conitzer
J. Zico Kolter
Kevin Leyton-Brown (University of British Columbia)

In Person and Zoom Participation.  See announcement.

Add event to Google
Add event to iCal