Please join the Department of Electrical and Computer Engineering for a seminar series featuring guest speaker Jinshuo Dong, a postdoctoral researcher in the Department of Computer Science
at Northwestern University, for a presentation titled, “The Design Choice of Privacy as a Rigid Constraint.” This seminar will take place on Tuesday, October 17, in Siegel Hall Auditorium (room 118) from 12:45–1:45 p.m. This event is open to the public.
Abstract
Privacy is a top-priority concern when modern data science achievements are applied to sensitive data. Typically, it is imposed as a hard constraint, while other measures, such as utility, are set as the optimization objective. In this talk, we delve into this design choice by analyzing its consequences in two specific scenarios. More specifically, we provide answers to the following questions:
1. Is it possible to deplete the privacy budget in the most basic application of differential privacy?
2. Can we battle against strategic behavior while maintaining hard privacy constraints in legal proceedings (in particular, electronic discovery)?
The first question suggests a hypothesis testing perspective of differential privacy, and for the second we design an incentive-compatible mechanism that relates the privacy loss to a geometric quantity—the number of extremal points after a projective transformation. Our findings suggest that privacy as a rigid constraint warrants careful consideration depending on the context. (Based on joint works with Aaron Roth, Weijie Su, Linjun Zhang, Liren Shan, Jason Hartline, and Aravindan Vijayaraghavan).
Bio
Jinshuo Dong is a postdoctoral researcher in the Department of Computer Science at Northwestern University. He earned his Ph.D. in applied mathematics from the University of Pennsylvania, supervised by Aaron Roth. Before that, he received his bachelor’s degree in mathematics from Peking University in 2014. His research revolves around data privacy and integrates various tools from statistics and applied mathematics into the subject.