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:Partial least squares for dependent data - Tatyana
Krivobokova (Georg-August-Universität Göttingen)
DTSTART;TZID=Europe/London:20180510T110000
DTEND;TZID=Europe/London:20180510T120000
UID:TALK105403AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/105403
DESCRIPTION:We consider the linear and kernel partial le
ast squares algorithms for dependent data and stud
y the consequences of ignoring the dependence both
theoretically and numerically. For linear partial
least squares estimator we derive convergence rat
es and show that ignoring non-stationary dependenc
e structures can lead to inconsistent estimation.
For kernel partial least squares estimator we esta
blish convergence rates under a source and an effe
ctive dimensionality conditions. It is shown both
theoretically and in simulations that long range d
ependence results in slower convergence rates. A p
rotein dynamics example illustrates our results an
d shows high predictive power of partial least squ
ares.
This is joint work with Marco Singer\, A
xel Munk and Bert de Groot.
LOCATION:Seminar Room 2\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR