•   When: Thursday, April 20, 2023 from 11:00 AM to 12:00 PM
•   Speakers: Daniel Schoepflin
•   Location: ENGR 4201
•   Export to iCal

 Abstract:  When designing algorithms in multi-agent systems involving self-interested agents, it is crucial that one consider the incentives of the agents in the design of the algorithm so that the algorithm may arrive at good outcomes.  The field of algorithmic mechanism design utilizes tools from game theory and economics to achieve just this goal, i.e., algorithms which are robust to strategic behavior.  An expansive literature has grown from this field, and this literature spans computer science, economics, and operations research and has proposed many beautiful and foundational theoretical results. Unfortunately, empirical literature in economics has demonstrated that the foundational assumption in much of this literature – that agents are perfectly rational – is unfounded in practice.  To address this gap between theory and practice, it is necessary to design mechanisms for “real-world” agents, by relaxing the assumption that agents can reason perfectly.

In this talk, I will revisit the canonical and important mechanism design problem of auction design with real-world agents in mind.  I will introduce an emerging class of auctions known as (deferred acceptance) clock auctions, argue why they are much better suited for practical applications than traditional “sealed-bid” auctions, and give a survey of my thesis results which dramatically expand our understanding of this important class of mechanisms by analyzing them from the traditional computer science perspective of algorithm design and worst-case analysis.

Bio:  Daniel Schoepflin is a doctoral student in the Computer Science Department of Drexel University advised by Dr. Vasilis Gkatzelis.  His main research interests are algorithmic mechanism design and approximation algorithms with a particular focus on designing and analyzing practical mechanisms and bridging the gap between theory and practice.  In Fall 2019 he was a visiting doctoral student at the Simons Institute for the Theory of Computing at UC Berkeley for the Market Design program and in Summer 2022 he was a Research Intern at Google Research hosted by Dr. Gagan Goel and the Google Ad Auctions team.