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:Statistics Reading Group
SUMMARY:An Introduction to Variational Methods for Approxi
mate Inference in Graphical Models - Silvia Chiapp
a
DTSTART;TZID=Europe/London:20100217T170000
DTEND;TZID=Europe/London:20100217T180000
UID:TALK22381AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/22381
DESCRIPTION:Many graphical models of practical interest do not
admit exact probabilistic\ninference and therefor
e require the use of approximations. Variational\n
methods are deterministic approximation methods th
at have been extensively\nstudied and applied in t
he Machine Learning community as an alternative to
\nstochastic approximation methods like Markov Cha
in Monte Carlo (MCMC)\nmethods. Some of the charac
teristics that make variational methods more\nattr
active than MCMC methods are their often preferabl
e computational cost\nand their ability to provide
bounds on distributions. Whilst in theory they\nc
annot generate exact results since they are based
on an analytical\napproximation to the distributio
n of interest\, they have proven to give\nsimilar
or superior performance to MCMC methods in severa
l real-world\napplications. In this talk I will ex
plain through simple examples the basic\nprinciple
s and properties of variational methods and presen
t some successful\napplications. I will also show
how loopy belief propagation can be\nformulated in
a variational framework and introduce a few exten
sions that\nhave been derived using this viewpoint
. I will finally describe the link\nbetween variat
ional transformations and convex duality.\n
LOCATION:MR5\, CMS
CONTACT:Richard Samworth
END:VEVENT
END:VCALENDAR