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:Surrogate models in Bayesian Inverse Problems - Ar
etha Teckentrup (University of Edinburgh)
DTSTART;TZID=Europe/London:20180208T113000
DTEND;TZID=Europe/London:20180208T123000
UID:TALK100207AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/100207
DESCRIPTION:Co-authors: Andrew Stuart (Caltech) \, H
an Cheng Lie and Timm Sullivan (Free University Be
rlin)
We are interested in
the inverse problem of estimating unknown paramet
ers in a mathematical model from observed data. We
follow the Bayesian approach\, in which the solut
ion to the inverse problem is the probability dist
ribution of the unknown parameters conditioned on
the observed data\, the so-called posterior distri
bution. We are particularly interested in the case
where the mathematical model is non-linear and ex
pensive to simulate\, for example given by a parti
al differential equation. We consider the use of s
urrogate models to approximate the Bayesian poster
ior distribution. We present a general framework f
or the analysis of the error introduced in the pos
terior distribution\, and discuss particular examp
les of surrogate models such as Gaussian process e
mulators and randomised misfit approaches.
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR