BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Particle filters and curse of dimensionality - Patrick Rebeschini 
 (Princeton)
DTSTART:20140221T140000Z
DTEND:20140221T150000Z
UID:TALK51053@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:A problem that arises in many applications is to compute the c
 onditional distributions of stochastic models given observed data. While e
 xact computations are rarely possible\, particle filtering algorithms have
  proved to be very useful for approximating such conditional distributions
 . Unfortunately\, the approximation error of particle filters grows expone
 ntially with dimension\, a phenomenon known as curse of dimensionality. Th
 is fact has rendered particle filters of limited use in complex data assim
 ilation problems that arise\, for example\, in weather forecasting or mult
 i-target tracking. In this talk I will show that it is possible to develop
  “local” particle filtering algorithms whose approximation error is di
 mension-free. By exploiting conditional decay of correlations properties o
 f high-dimensional models\, we prove for the simplest possible algorithm o
 f this type an error bound that is uniform both in time and in the model d
 imension. (Joint work with R. van Handel)
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR