BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Applied and Computational Analysis
SUMMARY:Inertial primal dual splitting methods - Dirk Lore
nz (Technische Universität Braunschweig)
DTSTART;TZID=Europe/London:20131010T150000
DTEND;TZID=Europe/London:20131010T160000
UID:TALK44385AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/44385
DESCRIPTION:Many problems in mathematical imaging can be phras
ed as convex minimization problems or convex-conca
ve saddle point problems. In both cases\, the resp
ective optimality system is an inclusion with a (m
aximally) monotone operator. To solve these inclus
ions\, several splitting methods have been propose
d which rely on the idea that the monotone operato
r can be split up into simpler parts for which\, e
.g. the resolvent can be applied easily.\n\nIn thi
s talk we present a fairly general splitting metho
d which works for inclusions in which the operator
can be split such that one operator is co-coerciv
e and for the other a certain preconditioned resol
vent is easily applicable. We arrive at an inertia
l forward backward splitting method for which we p
rove weak convergence under fairly general assumpt
ions. It is show that the methods covers several e
xisting methods such as Polyak's heavy ball method
\, Nesterov's accelerated gradient descent\, the f
orward-backward splitting method and Beck and Tebo
ulle's FISTA. We illustrate the applicability and
performance on numerous problems such as the Rudin
-Osher-Fatemi denoising and deconvolution or the O
sher-Sole-Vese denoising.\n\nThis is joint work wi
th Thomas Pock (TU Graz).\n
LOCATION:MR 14\, CMS
CONTACT:Carola-Bibiane Schoenlieb
END:VEVENT
END:VCALENDAR