BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Classification of fat-tailed clusters in high dimensions
DTSTART:20230606T120000Z
DTEND:20230606T130000Z
DTSTAMP:20230924T182900Z
UID:indico-event-9837@indico.math.cnrs.fr
DESCRIPTION:Speakers: Gabriele Sicuro (King's College\, London.)\n\nAbstra
ct:I will discuss the problem of learning a mixture of two clouds of data
points with generic centroids via empirical risk minimisation in the high
dimensional regime\, under the assumptions of generic convex loss and conv
ex regularisation. Each cloud of data points is obtained by sampling from
a possibly uncountable superposition of Gaussian distributions\, whose var
iance has a generic probability density ϱ. Our analysis covers therefore
a large family of data distributions\, including the case of power-law-tai
led distributions with no covariance. We study the generalisation performa
nce of the obtained estimator\, we analyse the role of regularisation and
the dependence of the separability transition on the distribution scale pa
rameters.\n\nhttps://indico.math.cnrs.fr/event/9837/
URL:https://indico.math.cnrs.fr/event/9837/
END:VEVENT
END:VCALENDAR