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 Reading Group @ CUED
SUMMARY:Topics in Expectation Propagation - Yingzhen Li (U
 niversity of Cambridge)\, Rich Turner
DTSTART;TZID=Europe/London:20160428T143000
DTEND;TZID=Europe/London:20160428T160000
UID:TALK65403AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/65403
DESCRIPTION:Approximate inference is key to modern probabilist
 ic modeling\, since exact inference/learning is in
 tractable for many models used in real-world appli
 cations. In this talk we present expectation propa
 gation (EP) as a general framework for fast and ac
 curate approximate inference. We give a wide range
  of applications of EP for both posterior approxim
 ation and marginal inference. We also show the fle
 xibility of algorithm design within the EP framewo
 rk by introducing factor graphs\, approximate dist
 ribution families and projection operators. Finall
 y we provide a justification of EP from a variatio
 nal viewpoint and connect it to the Bethe free ene
 rgy approximation.\n
LOCATION:Engineering Department\, CBL Room 438
CONTACT:Yingzhen Li
END:VEVENT
END:VCALENDAR