2021 (modified: 25 Feb 2022)ALT 2021Readers: Everyone
Abstract:We provide sample complexity upper bounds for agnostically learning multivariate Gaussians under the constraint of approximate differential privacy. These are the first finite sample upper bounds f...