DocumentCode
3183874
Title
Modified back propagation algorithm for learning artificial neural networks
Author
Ahmed, Waleed A Maguid ; M. Saad, El ; Aziz, E.S.A.
Author_Institution
Cairo Univ., Egypt
Volume
1
fYear
2001
fDate
2001
Firstpage
345
Abstract
Back Propagation is now the most widely used tool in tile field of artificial neural networks. Many attempts try to enhance this algorithm to get minimum mean square error, less training time and small number of epochs. This paper first reviews the disadvantages of the Back Propagation algorithm. Next, the new modified back propagation is explained. Finally, comparison between the two algorithms is made through many examples
Keywords
backpropagation; neural nets; software performance evaluation; backpropagation algorithm; character recognition; convergence; function approximation; learning artificial neural networks; minimum mean square error; training time; Acoustic propagation; Algorithm design and analysis; Approximation algorithms; Artificial neural networks; Costs; Mean square error methods; Multi-layer neural network; Neural networks; Nonhomogeneous media; Stochastic processes;
fLanguage
English
Publisher
ieee
Conference_Titel
Radio Science Conference, 2001. NRSC 2001. Proceedings of the Eighteenth National
Conference_Location
Mansoura
Print_ISBN
977-5031-68-0
Type
conf
DOI
10.1109/NRSC.2001.929244
Filename
929244
Link To Document