Title :
An EM algorithm for singular state space models
Author_Institution :
Sch. of Electr. Eng., New South Wales Univ., Sydney, NSW, Australia
Abstract :
We develop a state space EM algorithm for the case when the state innovations covariance matrix is singular. Previous state space algorithms required this to be of full rank.
Keywords :
covariance matrices; optimisation; state-space methods; EM algorithm; covariance matrix; expectation maximization algorithm; singular state space models; Australia; Context modeling; Convergence; Covariance matrix; Maximum likelihood estimation; Parameter estimation; Perturbation methods; State-space methods; Technological innovation; White noise;
Conference_Titel :
Decision and Control, 2003. Proceedings. 42nd IEEE Conference on
Print_ISBN :
0-7803-7924-1
DOI :
10.1109/CDC.2003.1271681
