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:Signal Processing and Communications Lab Seminars
SUMMARY:Shrinkage Estimation in High Dimensions - Dr K. Pa
van Srinath\, CUED
DTSTART;TZID=Europe/London:20160602T150000
DTEND;TZID=Europe/London:20160602T160000
UID:TALK65819AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/65819
DESCRIPTION:We consider the problem of estimating a high-dimen
sional vector of parameters from a noisy one-time
observation. The noise vector is iid Gaussian with
known variance\, and the performance of the estim
ator is measured via squared-error loss. For this
problem\, _shrinkage estimators_\, which shrink th
e observed data towards a point or a target subspa
ce\, have evoked a lot of interest because they do
minate the simple maximum-likelihood estimator (wh
en the number of dimensions exceeds two). \n\nIn t
his talk\, we first review the key aspects of shri
nkage estimation\, and then introduce shrinkage es
timators that use the data to determine a "good" t
arget subspace to shrink the data towards. We give
concentration results for the squared-error loss
and convergence results for the risk of the propos
ed estimators. We also present simulation results
that validate the theory. \n\nThis is joint work w
ith Ramji Venkataramanan.
LOCATION:LR6\, Department of Engineering
CONTACT:Dr Ramji Venkataramanan
END:VEVENT
END:VCALENDAR