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:Machine Learning @ CUED
SUMMARY:Convex Variational Bayesian Inference for Large Sc
 ale Generalized Linear Models - Hannes Nickisch
DTSTART;TZID=Europe/London:20090827T110000
DTEND;TZID=Europe/London:20090827T120000
UID:TALK19664AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/19664
DESCRIPTION:Bayesian inference for most generalized linear mod
 els is analytically not tractable. We show that a 
 well known variational relaxation leads to a conve
 x problem for any log-concave model and provide a 
 generic double loop algorithm for solving it on mo
 dels with arbitrary super-Gaussian potentials. We 
 iteratively decouple the criterion\, so that most 
 of the computational work is done by solving large
  linear systems\, rendering our algorithm much fas
 ter than previously proposed solvers. We evaluate 
 our method on problems of Bayesian active learning
  for large binary classification models.
LOCATION:Engineering Department\, CBL Room 438
CONTACT:Carl Edward Rasmussen
END:VEVENT
END:VCALENDAR