Teresa Anna Steiner (DTU) will give a talk on “Differential Privacy for Dynamic Data“.

When: Friday, March 17, 2023, 11:00 am

Where: Seminar room I, Jakob–Haringer–Straße 2A (JAK2A OG 1.33)

Abstract: Differential privacy is a rigorous privacy standard widely used in data analysis. Informally, a randomized algorithm is differentially private, if its output distribution does not depend too much on any individual data point in the input (which consists of many such data points). The goal in differential privacy is to design algorithms which are provably differentially private, while still giving accurate results. From a theoretical point of view, the interesting question is to find the best possible privacy-accuracy tradeoff for any given query.

In most real-world applications, the data we want to analyze changes over time. If we keep answering queries on an evolving data set, the privacy guarantees of standard differentially private algorithms significantly weaken over time. The goal of the research field of “differential privacy under continual observation” is to design algorithms which remain provably private even in the dynamic setting. In this talk, I will give an introduction to differential privacy under continual observation, and briefly talk about a recently submitted result about maintainingdifferentially private dynamic histograms.