Title :
Optimal matching-score network with growth capability
Author :
Leung, W.F. ; Luk, Andrew
Author_Institution :
Dept. of Electron. Eng., City Polytech. Of Hong Kong, Kowloon, Hong Kong
Abstract :
A robust and optimal matching-score network is presented. The algorithm is based on a steepest descent training procedure. The net is capable of learning new patterns without recomputing the entire interconnection weight matrices. Simulation results are given to illustrate the strengths of the proposed design method.
Keywords :
feedforward neural nets; learning (artificial intelligence); growth capability; interconnection weight matrices; optimal matching-score network; steepest descent training procedure;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:19930989
