Speaker: Hannah Keller

Personal data is increasingly used in algorithms and decision-making across many aspects of our lives. Differential privacy provides a mathematical framework for balancing useful data analysis with the protection of individual privacy. It has been deployed in practice by organisations including Google, Apple, and the United States Census Bureau. Informally, differential privacy requires that the outcome of an analysis should not depend too strongly on whether any one individual’s data is included. This helps defend against attacks that try to infer membership in a dataset or reconstruct sensitive information about individuals. 

This talk will introduce the basic definition of differential privacy, explain its motivation, and give examples of algorithms that satisfy it.