Institute of Mathematics and Statistics,
University of Kent at Canterbury,
Statistics Seminar -- Trinity Term 1999
Thursday, 1st July 1999
2:00pm Dr Simon Godsill
(Cambridge University)
Title: Improvement strategies for Bayesian filtering
using particle filters
Abstract:
A topic of increasing importance in statistical modelling of large time
series datasets is sequential Monte Carlo computation of
posterior distributions (Bayesian filtering) using particle filters.
Recent advances in the area have opened up new possibilities
for models with hyperparameters and high-dimensional dynamic parameters.
We investigate here improvement strategies for
particle filters, using fixed-lag simulation techniques, bridging
distributions and Markov chain resampling ideas. The techniques
are compared and demonstrated by application to state space time series
models drawn from the audio and communications fields.
****************
Further details of the seminar programme and abstracts of
forthcoming talks are available from:
http://www.ukc.ac.uk/IMS/statistics/people/T.Sapatinas/Seminars/
seminars.html
All are very welcome. The Statistics Seminars will be held in
Mathematics Lecture Theatre in the Institute of Mathematics and
Statistics, Cornwallis Building.
The following WWW page contain details of how to get to the campus,
and how to find your way around:
http://www.ukc.ac.uk/ukc/location.html
================================================================
Dr Theofanis Sapatinas,
Lecturer in Statistics,
Postal Address: Institute of Mathematics and Statistics,
University of Kent at Canterbury,
Canterbury, Kent CT2 7NF,
United Kingdom.
Tel: ++(44)-1227-827253
Fax: ++(44)-1227-827932
Email: [log in to unmask]
URL: http://www.ukc.ac.uk/IMS/statistics/people/T.Sapatinas/
================================================================
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|