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:MRC Biostatistics Unit Seminars
SUMMARY:Inference for binary Markov random fields without
tears (or MCMC). - Nial Friel\, University of Glas
gow
DTSTART;TZID=Europe/London:20070605T143000
DTEND;TZID=Europe/London:20070605T153000
UID:TALK6779AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/6779
DESCRIPTION:Binary Markov random fields have played an importa
nt role in the development of MCMC methods. The Me
tropolis algorithm was designed to sample from the
Ising model\, for example. In this talk we will p
resent an non-MCMC based approach to carrying out
inference for such models. Here I will present an
algorithm which allows\, for example\, direct samp
ling from the Markov random field\, computation of
marginal distributions of lattice points\, but cr
ucially which also allows inference for the parame
ters of the MRF. These algorithms can be carried o
ut exactly if the dimension of the lattice is of m
oderate size. We present an approximate inference
scheme for the case where the lattice is of large
dimension. This is joint work with Havard Rue (NTN
U\, Trondheim).
LOCATION:Large Seminar Room\, 1st Floor\, Institute of Publ
ic Health\, University Forvie Site\, Robinson Way\
, Cambridge
CONTACT:Nikolaos Demiris
END:VEVENT
END:VCALENDAR