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:Isaac Newton Institute Seminar Series
SUMMARY:Asymptotics for ABC algorithms - Judith Rousseau
(University of Oxford\; Université Paris-Dauphine)
DTSTART;TZID=Europe/London:20180118T140000
DTEND;TZID=Europe/London:20180118T144500
UID:TALK97789AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/97789
DESCRIPTION:Approximate Bayesoan Computation algorithms (ABC)
are used in cases where the likelihood is intract
able. To simulate from the (approximate) posterio
r distribution a possiblity is to sample new data
from the model and check is these new data are clo
se in some sense to the true data. The output of t
his algorithms thus depends on how we define the n
otion of closeness\, which is based on a choice of
summary statistics and on a threshold. Inthis wor
k we study the behaviour of the algorithm under th
e assumption that the summary statistics are conce
ntrating on some deterministic quantity and charac
terize the asymptotic behaviour of the resulting a
pproximate posterior distribution in terms of the
threshold and the rate of concentration of the sum
mary statistics. The case of misspecified models i
s also treated where we show that surprising asymp
totic behaviour appears.
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR