Title :
A redundancy approach to classifier training
Author :
Al-Alaoui, Mohamad Adnan ; Mouci, Rodolphe ; Mansour, Mohamad
Author_Institution :
Dept. of Electr. & Comput. Eng., American Univ. of Beirut, Lebanon
Abstract :
The Al-Alaoui algorithm is a weighted mean-square-error (MSE) approach to pattern recognition. It employs redundancy, reintroducing the erroneously classified samples to increase the population of their corresponding classes. The algorithm was originally developed for single-layer neural networks. In this paper the algorithm is extended to multilayer neural networks. It is also shown that the application of the Al-Alaoui algorithm to multilayer neural networks speeds up the convergence of the backpropagation algorithm. The application of the Al-Alaoui algorithm to the Levenberg-Marquardt algorithm for difficult pattern classification problems reduces the number of patterns that are erroneously classified
Keywords :
backpropagation; mean square error methods; multilayer perceptrons; pattern classification; redundancy; Al-Alaoui algorithm; Levenberg-Marquardt algorithm; backpropagation algorithm; classifier training; erroneously classified samples; multilayer neural networks; pattern classification problems; pattern recognition; redundancy; weighted mean-square-error approach; Backpropagation algorithms; Convergence; Multi-layer neural network; Neural networks; Neurons; Pattern classification; Pattern recognition;
Conference_Titel :
Electronics, Circuits and Systems, 2000. ICECS 2000. The 7th IEEE International Conference on
Conference_Location :
Jounieh
Print_ISBN :
0-7803-6542-9
DOI :
10.1109/ICECS.2000.913033
