BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium : New Results at the Crossroads of Convexity\, Learn
 ing and Information Theory
DTSTART:20161219T161500
DTEND:20161219T173000
DTSTAMP:20260916T014015Z
UID:dac8f5fe472aa0dae262596db87e5b094906d9c3901cac98338a404c
CATEGORIES:Conferences - Seminars
DESCRIPTION:By : Sébastien Bubeck - Microsoft Research\n\nVideo of his ta
 lk\n\nAbstract :\nI will present three new results (no background in optim
 ization will be assumed\, all concepts will be defined and motivated): (i)
  the Cramer transform of the uniform measure on a convex body is a univers
 al self-concordant barrier\; (ii) projected gradient descent with Gaussian
  noise allows to sample from a log-concave measure in polynomial time\; an
 d (iii) Thompson sampling combined with a multi-scale exploration solves t
 he Bayesian convex bandit problem. The unifying theme in these results is 
 the interplay between concepts from convex geometry\, learning and informa
 tion theory.\nJoint work with Ronen Eldan.\n\nBio :\nSébastien Bubeck is
  a Researcher in the Theory Group at Microsoft Research Redmond. Prior to 
 MSR\, he was an Assistant Professor in the Department of Operations Resea
 rch and Financial Engineering at Princeton University. He received his MS 
 in Mathematics from the Ecole Normale Supérieure de Cachan and his PhD fr
 om the University of Lille 1\, France\, where his PhD thesis won several p
 rizes including the Jacques Neveu prize for the best French PhD thesis in 
 Probability and Statistics. He was awarded the Sloan Research Fellowship i
 n Computer Science in 2015\, the Best Student Paper Award at COLT (Confere
 nce on Learning Theory) 2009 and the Best Paper Award at COLT 2016. He is 
 also the author of two monographs published in Foundations and Trends in M
 achine Learning\, “Regret Analysis of Stochastic and Nonstochastic Multi
 -armed Bandit Problems”\, and “Convex Optimization: Algorithms and Com
 plexity”.\n\nMore information
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR
