Title :
Improvement of convergence properties of Wang neural network
Author_Institution :
Inst. of Radio Electron., Tech. Univ. of Brno
fDate :
10/27/1994 12:00:00 AM
Abstract :
A method for improving the convergence properties of the Wang neural network is presented. The original Wang network is modified and completed by the Kalman gain estimator that suboptimally controls the convergence step of the settling process. The performances of the original Wang network and the modified network are compared via an illustrative example
Keywords :
Kalman filters; analogue processing circuits; convergence; neural nets; Kalman gain estimator; Wang neural network; convergence properties; modified network; settling process; suboptimal control;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:19941273
