CCI's Computer Science Departmental Talk Series Presents: Bo Waggoner

 

Abstract: This talk will be based around a simple setting: a person (or algorithm) makes a prediction, and then a "scoring rule" (or loss function) is used to evaluate its accuracy. What do we mean by "prediction" here, and what makes a prediction good? Waggoner will attempt to share his fascination both with the mathematical depths of these questions (including connections to geometry and information theory) and with exciting applications in game theory and machine learning. By the end, our goal will be to use this tool to design theoretically-grounded "markets for data".

 

About the Speaker: Bo Waggoner is a postdoctoral fellow at the University of Pennsylvania's Warren Center for Network and Data Sciences. His work focuses on systems for learning and aggregating information in contexts with strategic behavior, privacy, or fairness considerations. He received his PhD from Harvard in 2016. 

 

RSVP at drexelcci.info/waggoner