BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Pravesh Kothari and Ankur Moitra (CMU and MIT) DTSTART:20210609T170000Z DTEND:20210609T183000Z DTSTAMP:20260423T021012Z UID:TCSPlus/28 DESCRIPTION:Title: Robustly Learning Mixtures of Gaussians\nby Pravesh Kothari and Ankur Moitra (CMU and MIT) as part of TCS+\n\n\nAbstract\nFor a while now the p roblem of robustly learning a high-dimensional mixture of Gaussians has ha d a target on its back. The first works in algorithmic robust statistics g ave provably robust algorithms for learning a single Gaussian. Since then there has been steady progress\, including algorithms for robustly learnin g mixtures of spherical Gaussians\, mixtures of Gaussians under separation conditions\, and arbitrary mixtures of two Gaussians. In this talk we wil l discuss two recent works that essentially resolve the general problem. T here are important differences in their techniques\, setup\, and overall q uantitative guarantees\, which we will discuss.\n\nThe talk will cover the following independent works:\n- Liu\, Moitra\, "Settling the Robust Learn ability of Mixtures of Gaussians"\n- Bakshi\, Diakonikolas\, Jia\, Kane\, Kothari\, Vempala\, "Robustly Learning Mixtures of $k$ Arbitrary Gaussians "\n LOCATION:https://researchseminars.org/talk/TCSPlus/28/ END:VEVENT END:VCALENDAR