BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Detecting Change Points in Multidimensional Functional Data - John
  Aston\, Warwick
DTSTART:20110304T160000Z
DTEND:20110304T170000Z
UID:TALK28564@talks.cam.ac.uk
CONTACT:Richard Nickl
DESCRIPTION:Change point detection in sequences of functional data is exam
 ined where the\nfunctional observations are dependent and where the distri
 butions of change\npoints from multiple subjects is required. Of particula
 r interest is the\ncase where the change point is an epidemic change (a ch
 ange occurs and then\nthe observations return to baseline at a later time)
 . The special case where\nthe covariance can be decomposed as a tensor pro
 duct is considered with\nparticular attention to the power analysis for de
 tection. This is of\ninterest in the application to functional magnetic re
 sonance imaging (fMRI)\,\nwhere the estimation of a full covariance struct
 ure for the\nthree-dimensional image is not computationally feasible. It i
 s found that\nuse of basis projections such as principal components for de
 tection of the\nchange points can be optimal in situations where PCA is tr
 aditionally\nthought to perform badly.\n[Joint work with Claudia Kirch\, K
 arlsruhe Institute of Technology]\n\n\nhttp://www2.warwick.ac.uk/fac/sci/s
 tatistics/staff/academic-research/aston/
LOCATION:MR12\, CMS\, Wilberforce Road\, Cambridge\, CB3 0WB
END:VEVENT
END:VCALENDAR