BEGIN:VCALENDAR
VERSION:2.0
PRODID:researchseminars.org
CALSCALE:GREGORIAN
X-WR-CALNAME:researchseminars.org
BEGIN:VEVENT
SUMMARY:Gabriele Sicuro (King's College London\, United Kingdom)
DTSTART:20230606T120000Z
DTEND:20230606T130000Z
DTSTAMP:20260710T043814Z
UID:SeedSeminar/26
DESCRIPTION:Title: <a href="https://researchseminars.org/talk/SeedSeminar/
 26/">Classification of fat-tailed clusters in high dimensions</a>\nby Gabr
 iele Sicuro (King's College London\, United Kingdom) as part of Seed Semin
 ar of Mathematics and Physics\n\n\nAbstract\nI will discuss the problem of
  learning a mixture of two clouds of data points with generic centroids vi
 a empirical risk minimisation in the high dimensional regime\, under the a
 ssumptions of generic convex loss and convex regularisation. Each cloud of
  data points is obtained by sampling from a possibly uncountable superposi
 tion of Gaussian distributions\, whose variance has a generic probability 
 density ϱ. Our analysis covers therefore a large family of data distribut
 ions\, including the case of power-law-tailed distributions with no covari
 ance. We study the generalisation performance of the obtained estimator\, 
 we analyse the role of regularisation and the dependence of the separabili
 ty transition on the distribution scale parameters.\n
LOCATION:https://researchseminars.org/talk/SeedSeminar/26/
END:VEVENT
END:VCALENDAR