Computer Science Thesis Proposal

Thursday, April 20, 2017 - 1:00pm to 2:00pm


Traffic21 Classroom 6501 Gates Hillman Centers



The rise in social and economical interactions has presented the world with new pressing challenges, and the research community with a responsibility to develop a foundation for addressing them. Ubiquitous tools such as machine learning and optimization are already being used to address some of these challenges. However, many of the types of problems we would like to use machine learning and optimization to solve today do not fall into their classical frameworks. These include learning to design optimal auctions in an ever changing market, learning about agents by observing their interactions, learning from a large crowd, and optimizing the outcome of multi-agent mechanisms with an eye towards individual participants. Not accounting for complexities that arise from social and economical problems can have negative implications, such as developing solutions that look good in the classical setting but are ineffective in practice. In this thesis, my goal is to develop theoretical foundations for addressing some of these social and economic challenges in machine learning and optimization. This proposal specifically focuses on four of these challenges: Adaptive auction design and oracle-efficient learning, learning and optimization in Stackelberg games, learning from the crowd, and incentives and optimization in kidney exchange. Thesis Committee: Avrim Blum (Co-Chair) Ariel Procaccia (Co-Chair) Maria-Florina Balcan Tim Roughgarden (Stanford University) Robert Schapire (Microsoft Research, NYC) Copy of Thesis Summary 

For More Information, Contact:


Thesis Proposal