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:Bayesian optimization: A framework for optimal com
putational effort for experimental design - Winter
fors\, E (Ruprecht-Karls-Universitt Heidelberg)
DTSTART;TZID=Europe/London:20110719T170000
DTEND;TZID=Europe/London:20110719T173000
UID:TALK32092AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/32092
DESCRIPTION:DOE on models involving time or space dynamics is
often very computationally demanding. Predicting a
single experimental outcome may require significa
nt computation\, let alone evaluating a design cri
terion and optimizing it with respect to design p
arameters. To find the exact optimum of the design
criterion would typically take infinite computati
on\, and any finite computation will yield a resul
t possessing some uncertainty (due to approximatio
n of the design criterion as well as stopping the
optimization procedure). Ideally\, one would like
to optimize not only the design criterion\, but al
so the way it is approximated and optimized in ord
er to get the largest likely improvement in the de
sign criterion relative to the computational effor
t spent. Using a Bayesian method for the optimizat
ion of the design criterion (not only for calculat
ing the design criterion) can accomplish such an o
ptimal trade-off between (computational) resources
spent planning the experiment and expected gain f
rom carrying it out. This talk will lay out the c
oncepts and theory necessary to perform a fully Ba
yesian optimization that maximizes the expected im
provement of the design criterion in relation the
computational effort spent.\n
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:Mustapha Amrani
END:VEVENT
END:VCALENDAR