BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Interpolation of Deterministic Simulator Outputs using a Gaussian 
 Process Model - Ranjan\, P (Acadia University)
DTSTART:20110909T080000Z
DTEND:20110909T083000Z
UID:TALK32736@talks.cam.ac.uk
CONTACT:Mustapha Amrani
DESCRIPTION:For many expensive deterministic computer simulators\, the out
 puts do not have replication error and the desired metamodel (or statistic
 al emulator) is an interpolator of the observed data. Realizations of Gaus
 sian spatial processes (GP) are commonly used to model such simulator outp
 uts. Fitting a GP model to n data points requires the computation of the i
 nverse and determinant of n x n correlation matrices\, R\, that are someti
 mes computationally unstable due to near-singularity of R. This happens if
  any pair of design points are very close together in the input space. The
  popular approach to overcome near-singularity is to introduce a small nug
 get (or jitter) parameter in the model that is estimated along with other 
 model parameters. The inclusion of a nugget in the model often causes unne
 cessary over-smoothing of the data. In this talk\, we present a lower boun
 d on the nugget that minimizes the over-smoothing and an iterative regular
 ization approach to construct a predictor th at further improves the inter
 polation accuracy. We also show that the proposed predictor converges to t
 he GP interpolator.\n
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR