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:Worms and Bugs
SUMMARY:Exact simulation-based Bayesian inference for epid
emic models - TJ McKinley\, Department of Veterina
ry Medicine
DTSTART;TZID=Europe/London:20111107T120000
DTEND;TZID=Europe/London:20111107T130000
UID:TALK33679AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/33679
DESCRIPTION:Inference in epidemic models poses many challenges
\, not least because of missing or unobserved data
. A powerful method for tackling some of these iss
ues is to use a Bayesian framework and data-augmen
ted Markov chain Monte Carlo fitting algorithms. H
owever\, these techniques can become computational
ly intensive for large-scale systems. An alternati
ve is to use pseudo-marginal algorithms (O'Neill e
t al.\, 2000\; Beaumont\, 2003\; Andrieu and Rober
ts\, 2009)\, which provide methods for estimating
both exact and approximate posterior distributions
for the parameters-of-interest based on importanc
e sample estimates generated from model simulation
s. When the observation process is deterministic\,
then this requires that the model simulations mat
ch the observed data exactly\, which can be proble
matic in highly stochastic systems without the ava
ilability of large amounts of computing power. We
present some methods for reducing stochasticity an
d improving computational efficiency for simulatio
ns of epidemic models\, by conditioning the simula
tions on the model and data. We illustrate these t
echniques on real data for a variety of model/data
combinations.
LOCATION:DD47\, Cripps Court\, Queens' College
CONTACT:Prof. Julia Gog
END:VEVENT
END:VCALENDAR