BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium : Beyond stochastic gradient descent for large-scale
  machine learning
DTSTART:20151026T161500
DTEND:20151026T173000
DTSTAMP:20260407T101200Z
UID:310fa47b016976c8ff5ce7a5d977680b01b33ad0c1c48bc2e0f60637
CATEGORIES:Conferences - Seminars
DESCRIPTION:By : Francis Bach - INRIAVideo of his talkAbstract :\nMany mac
 hine learning and statistics problems are traditionally cast as convex opt
 imization problems. A common difficulty in solving these problems is the s
 ize of the data\, where there are many observations ("large n") and each o
 f these is large ("large p"). In this setting\, online algorithms such as 
 stochastic gradient descent which pass over the data only once\, are usual
 ly preferred over batch algorithms\, which require multiple passes over th
 e data. Given n observations/iterations\, the optimal convergence rates of
  these algorithms are O(1/\\sqrt{n}) for general convex functions and reac
 hes O(1/n) for strongly-convex functions. In this talk\, I will show how t
 he smoothness of loss functions may be used to design novel simple algorit
 hms with improved behavior\, both in theory and practice: in the ideal inf
 inite-data setting\, an efficient novel Newton-based stochastic approximat
 ion algorithm leads to a convergence rate of O(1/n) without strong convexi
 ty assumptions. (joint work with Eric Moulines)Bio :\nFrancis Bach is a re
 searcher at INRIA\, leading since 2011 the SIERRA project-team\, which is 
 part of the Computer Science Department at Ecole Normale Superieure in Par
 is\, France. He completed his Ph.D. in Computer Science at U.C. Berkeley\,
  working with Professor Michael Jordan\, and spent two years in the Mathem
 atical Morphology group at Ecole des Mines de Paris\, then he joined the W
 ILLOW project-team at INRIA/Ecole Normale Superieure from 2007 to 2010. Fr
 ancis Bach is interested in statistical machine learning\, and especially 
 in graphical models\, sparse methods\, kernel-based learning\, convex opti
 mization vision and signal processing. He received a Starting Grant from t
 he European Research Council in 2009.More information
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR