[Ml-stat-talks] Fwd: PACM Colloquium 4/8/13

Philippe Rigollet rigollet at Princeton.EDU
Mon Apr 8 10:51:34 EDT 2013


Hi everyone,

today our very own Ramon van Handel will speak in the PACM colloquium.
The abstract below should talk to anyone with interests in statistics or probability.

See you there!
Philippe
--
Philippe Rigollet
www.princeton.edu/~rigollet<http://www.princeton.edu/~rigollet>





Begin forwarded message:

From: Valerie Marino <vmarino at math.princeton.edu<mailto:vmarino at math.princeton.edu>>
Subject: PACM Colloquium 4/8/13
Date: April 8, 2013 8:58:42 AM EDT
To: undisclosed-recipients:;

DATE: Monday,  April 8, 2013

PLACE: 214 Fine Hall

TIME: 4:30 pm

SPEAKER: Ramon van Handel - Princeton University - ORFE

TITLE: Filtering in high dimension

ABSTRACT:
A problem that arises in many applications is to compute the conditional
distributions of stochastic models given observed data. While exact
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
exponentially with dimension. This phenomenon has rendered particle
filters of limited use in complex data assimilation problems that arise,
for example, in weather forecasting or oceanography. In this talk, I will
argue that it should be possible, at least in theory, to develop "local"
particle filtering algorithms whose approximation error is dimension-free
by exploiting conditional decay of correlations properties of
high-dimensional models. As a proof of concept, we prove for the simplest
possible algorithm of this type an error bound that is uniform both in
time and in the model dimension.  (Joint work with P. Rebeschini)


Cookies will be served at 3:55 p.m. in 217A Fine Hall

--
Valerie Marino
Program Secretary
Program in Applied & Computational Mathematics
Tel: 609-258-3703
Fax: 609-258-1735

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/ml-stat-talks/attachments/20130408/87e23e60/attachment.htm>


More information about the Ml-stat-talks mailing list