Abstract: Today, data sharing is the cornerstone of many modern applications. A common concern in such data sharing pipelines is privacy: organizations are responsible for protecting the privacy of their data, whether it represents user data or enterprise trade secrets. In this talk, we will discuss challenges of navigating privacy-utility tradeoffs in two modern data sharing pipelines. (1) First, we will discuss emerging challenges related to learning large machine learning models from private, federated data. Classical approaches (namely, differentially-private federated learning, DP-FL) are difficult to scale to large models. We will explore the feasibility of replacing DP-FL  with centralized training over differentially private synthetic data. We will show that finetuning a model on DP synthetic data can perform similarly to DP-FL in downstream model performance, with order(s)-of-magnitude lower communication and computation.  (2) Second, we will explore techniques for hiding trade secrets--which we define as a function of a data distribution---in data sharing applications. We will propose a metric called statistic maximal leakage, which measures the leakage of a data release mechanism about a specific trade secret. We will demonstrate some desirable properties of this metric and computationally-efficient methods for calculating it, as well as mechanisms that satisfy the metric. Finally, we will demonstrate how to use these mechanisms to release a tabular dataset with strong privacy-utility tradeoffs.

Bio: Giulia Fanti is the Angel Jordan Associate Professor of Electrical and Computer Engineering at Carnegie Mellon University. Her research interests span the security, privacy, and efficiency of distributed systems. She is a two-time fellow of the World Economic Forum’s Global Future Council on Cybersecurity and a member of NIST’s Information Security and Privacy Advisory Board. She is a co-director of CyLab-Africa and the Initiative for Cryptocurrencies and Contracts. Her work has been recognized with several awards, including best paper awards, a Sloan Fellowship, an Intel Rising Star Faculty Award, and an ACM SIGMETRICS Rising Star Award. She obtained her Ph.D. in EECS from U.C. Berkeley and her B.S. in ECE from Olin College of Engineering.