BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Rao-Blackwellized Particle Smoothing for Conditionally Linear Gaus
 sian Models (NOTICE CHANGED TIME!) - Dr Simo Sarkka\, Biomedical Engineeri
 ng and Computer Science Dept\, Aalto University\, Finland
DTSTART:20111214T140000Z
DTEND:20111214T144500Z
UID:TALK34752@talks.cam.ac.uk
CONTACT:Rachel Fogg
DESCRIPTION:Although Monte Carlo based particle filters and smoothers can 
 be used for approximate inference in almost any kind of probabilistic stat
 e space models\, the required number of samples for a sufficient accuracy 
 can be high. The efficiency of sampling can be improved by Rao-Blackwelliz
 ation\, where part of the state is marginalized out in closed form\, and o
 nly the remaining part is sampled. Because the sampled space has a lower d
 imension\, fewer particles are required. In this talk I will discuss on Ra
 o-Blackwellization in the context of conditionally linear Gaussian models\
 , and present efficient Rao-Blackwellized versions of previously proposed 
 particle smoothers.\n\n
LOCATION: Cambridge University Engineering Department\, Lecture Room 6
END:VEVENT
END:VCALENDAR