Events

MIC Seminar: Optimal Differentially Private Sampling of Unbounded Gaussians

Speaker: Argyris Mouzakis

Location: 60 Fifth Avenue, Room Center for Data Science

Date: Thursday, January 29, 2026

We provide the first O_{α,ε,δ}(d)-sample algorithm for sampling from unbounded Gaussian distributions under the constraint of (ε, δ)-differential privacy. This is a quadratic improvement over previous results for the same problem, settling an open question of Ghazi, Hu, Kumar, and Manurangsi. Based on joint work with Valentio Iverson and Gautam Kamath, which has appeared in COLT 2025, and is available on Arxiv.

Bio: Argyris Mouzakis is a final year PhD student in Computer Science at the University of Waterloo, advised by Gautam Kamath. Previously, he earned his MEng in Electrical and Computer Engineering from the National Technical University of Athens. He has been a recipient of the Cheriton and Onassis Foundation scholarships for the support of his doctoral studies.