Approximation of Markovian models with non-constant parameters

Abstract

A generalization of the canonical correlation analysis approach has been developed for non-stationary process generated by Markovian models with non-constant parameters. This generalization, is then used to develop two model reduction (approximation) algorithms. 

Cite this paper

@article{Desai1984ApproximationOM, title={Approximation of Markovian models with non-constant parameters}, author={U. B. Desai and Saibal Banerjee and Sayfollah Kiaei}, journal={The 23rd IEEE Conference on Decision and Control}, year={1984}, pages={1642-1644} }