摘要：The problem of computing Kalman smoothed state estimates for linear (perhaps time varying) systems is by now a very well studied problem.  As a result, multiple solutions are available, with two of the most well known being the Rauch-Tung-Striebel and Bryson-Frazier recursions.  However, all existing methods impose certain assumptions on invertibility of associated covariance matrices or independence of state and measurement noise.  This talk presents a new method that computes smoothed state estimates under more general conditions.  The work is motivated by the need for Kalman Smoothing as part of the "E-step" in using the expectation-maximisation method for system identification purposes, and the talk will also discuss this link.

简历：Brett Ninness was born in 1963 in Singleton, Australia and received his BE, ME and Ph.D degrees in Electrical Engineering from the University of Newcastle, Australia in 1986, 1991 and 1994 respectively. 

His research interests are in the areas of system identification and stochastic signal processing, in which he has authored approximately one hundred papers in journals and conference proceedings.  He has served on the editorial boards of Automatica, IEEE Transactions on Automatic Control and has served as Editor in Chief for IET Control Theory and Applications.

He is currently Pro-Vice Chancellor of the Faculty of Engineering and Built Environment at the University of Newcastle, Australia.