Title :
A method for training feed forward neural network to be fault tolerant
Author :
Elsimary, H. ; Mashali, S. ; Shaheen, S.
Author_Institution :
Electron. Res. Inst., Cairo, Egypt
Abstract :
A method for training a feedforward neural network to be fault tolerant against weight perturbations is described. The measure for fault tolerance is the deviation of the network´s output after training, when each interconnection weight is perturbed, from that output without perturbation. In this method, an attempt is made to keep that deviation as low as possible. This measure is used because it can represent that kinds of error which arises when neural networks are implemented in hardware
Keywords :
fault tolerant computing; feedforward neural nets; learning (artificial intelligence); perturbation techniques; error representation; fault tolerance; feedforward neural network; interconnection weight; network output deviation; training; weight perturbations; Artificial neural networks; Backpropagation algorithms; Computer networks; Fault tolerance; Fault tolerant systems; Feedforward neural networks; Feeds; Neural network hardware; Neural networks; Predictive models;
Conference_Titel :
Virtual Reality Annual International Symposium, 1993., 1993 IEEE
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-1363-1
DOI :
10.1109/VRAIS.1993.380747
