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:Robust statistical decisions under model misspecif
ication by re-weighted Monte Carlo samplers - Holm
es\, C (University of Oxford)
DTSTART;TZID=Europe/London:20140424T091500
DTEND;TZID=Europe/London:20140424T101500
UID:TALK52167AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/52167
DESCRIPTION:Large complex data sets typically demand approxima
te models at some level of specification. In such
situations it is important for the analyst to exam
ine the robustness of conclusions to approximate p
redictions. Recent developments in optimal control
and econometrics have established formal methods
for linear quadratic state space models (see e.g.
Hansen and Sargent 2008) by considering the local-
minimax outcome within an information divergence (
Kullback-Leibler) neighbourhood around the approxi
mating model. Here we show how these approaches ca
n be extended to arbitrary probability models usin
g Monte Carlo methods. We derive theoretical resul
ts establishing the uniqueness of the Kullback-Lei
bler criteria\, as well as Bayesian non-parametric
methods to sample from the space of probability d
istributions within a fixed divergence constraint.
\n
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:Mustapha Amrani
END:VEVENT
END:VCALENDAR