C O L L O Q U I U M

---

Modified Foster-Liapounov functions and

their applications to asymptotic stability of

nonlinear filters in hidden Markov models

Professor Tze Leung Lai

Statistics Department, Stanford University, USA

C.V. Starr Visiting Professor at HKU

Abstract

A long-standing problem in optimal filtering theory is the asymptotic stability of nonlinear filters in hidden Markov models, generalizing classical stability results for the Kalman filter in linear state-space models. After a brief review of important developments in this problem during the past decade, we propose a new appraoch based on measure-valued Markov chains and certain modified Foster-Liapounov functions. Some applications of this asymptotic stability theory are also given.

Date:	December 8, 2005 (Thursday)
Time:	4:00 – 5:00pm
Place:	Room 517, Meng Wah Complex

Tea will be held in Room 516, Meng Wah Complex at 3:40pm

All are welcome