T&P Seminar – Privacy in Feedback Decision Systems

Details

Privacy in Feedback Decision Systems

Tuesday, August 23, 2022, at 12:50 pm
Location: MAE-A 303

Dr. Matt Hale
Assistant Professor
Department of Mechanical & Aerospace Engineering
University of Florida

Abstract

Engineering systems are gathering more data than ever. In applications ranging from smart power grids to traffic systems to connected homes, data is increasingly used to drive feedback loops for the control of such systems or for learning in them. Often this data is user data in the sense that it belongs to an end user or provides information about them, and increasingly this user data is sensitive. For example, the location of a user’s vehicle throughout the day can reveal their activities, close relationships, and places of work and residence. What emerges is a need to simultaneously safeguard sensitive data (for the benefit of users) and preserve that data’s usefulness in decision systems, namely those used in control and learning. In this talk, I will present recent work from my group that helps address this need by developing new privatization strategies for a class of symbolic systems, namely Markov chains, which are used in a wide variety of engineering applications. This work has made the first connections between symbolic systems and a formal privacy framework called “differential privacy,” and this talk will cover the novel, mathematically rigorous privacy protections that we have developed, along with formal guarantees that privatized data remains useful in feedback decision systems. I will also go over recent developments that bring privacy to several widely used control setups, including linear-quadratic control and multi-agent control.

Biography

Matthew Hale is an Assistant Professor of Mechanical and Aerospace Engineering at the University of Florida. He received his BSE in Electrical Engineering summa cum laude from the University of Pennsylvania in 2012 and his MS and Ph.D. in Electrical and Computer Engineering from the Georgia Institute of Technology in 2015 and 2017, respectively. His research interests include multi-agent systems, mobile robotics, privacy in control, artificial intelligence, and distributed optimization. He received an NSF CAREER Award in 2020, an ONR YIP in 2022, and a 2022 Excellence Award for Assistant Professors from the University of Florida (for being one of the top 10 assistant professors across all colleges on campus).

MAE Faculty Host: Carl Crane