BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Seminar by Prof. Anthony Man Cho So\, Chinese University of Hong K
 ong
DTSTART:20170622T120000
DTEND:20170622T133000
DTSTAMP:20260429T233210Z
UID:82a0ace39a5a66a9d63aa4611f78bd935c5ca6028782562a35cb1648
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Anthony Man-Cho So\, Chinese University of Hong Kong\n"E
 rror Bounds for Structured Optimization Problems and Convergence Analysis 
 of First-Order Methods in the Absence of Strong Convexity"\n\nAbstract\n\n
 In recent years\, we have witnessed a widespread use of first-order method
 s (FOMs) to solve large-scale structured convex optimization problems in m
 achine learning and signal processing. One fundamental issue in this line 
 of study is the convergence properties of the FOMs. Under strong convexity
  assumptions on the objective\, the linear convergence of various FOMs can
  be established in a fairly standard and straightforward manner. However\,
  such assumptions are not satisfied in most applications of interest. In t
 his talk\, we will present a framework for analyzing the convergence rates
  of FOMs. A key component of this framework is a so-called error bound con
 dition\, which stipulates that the distance from any candidate solution to
  the optimal solution set of the problem at hand can be estimated by certa
 in easily computable optimality measure. We will show that many structured
  convex optimization problems in machine learning satisfy the error bound 
 condition\, even though they do not have strongly convex objectives. Conse
 quently\, we are able to show that many FOMs have a linear rate of converg
 ence when applied to those problems. If time permits\, we shall also discu
 ss the application of our framework to certain structured non-convex optim
 ization problems.\n 
LOCATION:EPFL\, ODY 4.03\, VIP Room http://plan.epfl.ch/?zoom=19&recenter_
 y=5863800.12869&recenter_x=731560.22521&layerNodes=fonds\,batiments\,label
 s\,information\,parkings_publics\,arrets_metro\,transports_publics&floor=4
 &q=ODY_4.03
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR
