Robustly Learning Mixtures of Gaussians
Abstract: For a while now the problem of robustly learning a high-dimensional mixture of Gaussians has had a target on its back. The first works in algorithmic robust statistics gave provably robust algorithms for learning a single Gaussian. Since then there has been steady progress, including algorithms for robustly learning mixtures of spherical Gaussians, mixtures of Gaussians under separation conditions, and arbitrary mixtures of two Gaussians. In this talk we will discuss two recent works that essentially resolve the general problem. There are important differences in their techniques, setup, and overall quantitative guarantees, which we will discuss.
The talk will cover the following independent works:
- Liu, Moitra, "Settling the Robust Learnability of Mixtures of Gaussians"
- Bakshi, Diakonikolas, Jia, Kane, Kothari, Vempala, "Robustly Learning Mixtures of k Arbitrary Gaussians"
To watch the talk:
- Watching the live stream. At the announced start time of the talk (or a minute before), a live video stream will be available on our "next talk" page. Simply connect to the page and enjoy the talk. No webcam or registration is needed. Questions and comments during the talk are welcome (text only, unfortunately); simply post a comment below the live video stream on YouTube.
- Watching the recorded talk offline. The recorded talk will be made available shortly after the talk ends on our YouTube page. (Please leave a comment if you enjoyed it!)
Contact: Bonnie Leung firstname.lastname@example.org