BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An adaptive backtracking strategy for non-smooth composite optimis
 ation problems - Luca Calatroni (Ecole Polytechnique)
DTSTART:20180412T140000Z
DTEND:20180412T150000Z
UID:TALK103207@talks.cam.ac.uk
CONTACT:Carola-Bibiane Schoenlieb
DESCRIPTION:In this talk we present and backtracking strategy for a varian
 t of the Beck and Teboulle's Fast Iterative Shrinkage/Thresholding Algorit
 hm (FISTA) which has been recently proposed by Chambolle and Pock (2016) f
 or strongly convex objective functions. Differently from standard Armijo-t
 ype line searching\, our backtracking rule allows for local increase and d
 ecrease of the Lipschitz constant estimate along the iterations\, i.e. dec
 rease/increase of the gradient step size. For such adaptive strategy we pr
 ove accelerated convergence rates showing in particular linear convergence
  in smooth cases.  We validate the resulting algorithm on some exemplar im
 age denoising problems where strong convexity appears typically after smoo
 thing of the regularisation term.\n\nThis is joint work with A. Chambolle 
 (CMAP\, École Poltechnique).
LOCATION:MR14\, Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR