Bernard Lecture "Differential Privacy: Public Methods for Private Data" presented by Cynthia Dwork, Ph.D., Harvard University

Title: Differential Privacy: Public Methods for Private Data

Abstract: Differential privacy is a mathematically rigorous definition of privacy tailored to the statistical analysis of large datasets.  Simple aggregate statistics of confidential data always leak some information.  Each of those statistics is like a small x-ray:  An individual exposure causes little damage, but, just as risk from x-rays accumulates to cause serious physical harm, information leakage due to multiple statistics can in combination completely destroy privacy.  Differential privacy provides a quantitative measure of privacy loss and mathematics for how it accumulates.  Equipped with this measure, researchers and practitioners design methods to control privacy loss, just as algorithms experts and programmers design methods to minimize the running time of their programs. 

Differential Privacy is now widely deployed in industry.  In the public sphere, Differential Privacy most notably served as the foundation for the confidentiality protections legally required in the 2020 US Decennial Census. For the first time, this allowed complete transparency regarding the Census Bureau’s privacy-preserving methods, enabling the publication of a detailed description of the algorithm and the GitHub codebase.

This talk will explain the Differential Privacy guarantee, give some intuition about how to achieve it, and discuss its uses in industry and government.

Dr. Cynthia Dwork, Gordon McKay Professor of Computer Science at Harvard, is known for inventing differential privacy, non-malleable cryptography, and proofs of work; for launching the theory of algorithmic fairness; and for seminal contributions in lattice-based cryptography, distributed computing, and ensuring validity in adaptive data analysis.  Her many honors include the Japan Prize, the National Medal of Science, the Hamming Medal, and the Dijkstra, Gödel, Knuth, and Kanellakis Awards. She is a member of the US National Academy of Scences and the National Academy of Engineering, and a Fellow of the American Academy of Arts and Sciences and the American Philosophical Society.

Sponsored by the Department of Mathematics and Computers Science; and the Bernard Society in honor of the late Professor Emeritus Richard Bernard. Free and open to the Public.