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:Machine Learning @ CUED
SUMMARY:The Blended Paradigm: A Bayesian Approach to Hand
ling Outliers and Misspecified Models - Prof. Ste
ven MacEachern (Ohio State University)
DTSTART;TZID=Europe/London:20141215T100000
DTEND;TZID=Europe/London:20141215T110000
UID:TALK56674AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/56674
DESCRIPTION:Bayesian methods have proven themselves to be enor
mously successful across a wide range of scientifi
c problems\, with analyses ranging from the simple
one-sample problem to complicated hierarchical mo
dels. They have many well-documented advantages ov
er competing methods. However\, Bayesian methods r
un into difficulties for two major and prevalent c
lasses of problemsâ€”handling data sets with outlier
s and dealing with model misspecification. In both
cases\, standard Bayesian analyses fall prey to t
he hubris that is an integral part of the Bayesian
paradigm. The large sample behavior of the analys
is is driven by the likelihood. We propose the use
of restricted likelihood as a single solution to
both of these problems. When working with restrict
ed likelihood\, we summarize the data\, x\, throug
h a set of (insufficient) statistics T(x) and upda
te our prior distribution with the likelihood of T
(x) rather than the likelihood of x. By choice of
T(x)\, we retain the main benefits of Bayesian met
hods while reducing the sensitivity of the analysi
s to selected features of the data. The talk will
motivate the blended paradigm\, discuss properties
of the method and choice of T(x)\, cover the main
computational strategies for its implementation\,
and illustrate its benefits. This is joint work
with Yoonkyung Lee and John Lewis.
LOCATION:Engineering Department\, CBL Room BE-438.
CONTACT:Zoubin Ghahramani
END:VEVENT
END:VCALENDAR