Cryptography, Security and Privacy
P.4.4
Lecture
Information Privacy II

A Better Bound Gives a Hundred Rounds: Enhanced Privacy Guarantees via f-Divergences

Jiachun Liao, Lalitha Sankar, Oliver Kosut, Shahab Asoodeh, Flavio Calmon

Date & Time

01:00 am – 01:00 am

Abstract

We derive the optimal differential privacy (DP) parameters of a mechanism that satisfies a given level of Renyi differential privacy (RDP). Our result is based on the joint range of two f-divergences that underlie the approximate and the Renyi variations of differential privacy. We apply our result to the moments accountant framework for characterizing privacy guarantees of stochastic gradient descent. When compared to the state-of-the-art, our bounds may lead to about 100 more stochastic gradient descent iterations for training deep learning models for the same privacy budget.


Presenters

Jiachun Liao

Arizona State University

Lalitha Sankar

Arizona State University

Oliver Kosut

Arizona State University

Shahab Asoodeh

Harvard University

Flavio Calmon

Harvard University

Session Chair

Flavio Calmon

Harvard University