Title :
The back propagation method for CNN
Author :
Mizutani, Hikaru
Author_Institution :
Dept. of Electr. Eng., Shonan Inst. of Technol., Fujisawa, Japan
fDate :
30 May-2 Jun 1994
Abstract :
In this paper, a new back propagation method is presented. The method does not require a sigmoid function but a simple piecewise-linear function for the neurons. Because a cellular neural network (CNN) is able to use this kind of simple nonlinear function, the presented method may be applied to CNNs. Also the presented method can be applied not only to CNNs but also to any multilayer type neural network. The simple piecewise-linear function has the advantage that computing the function or fabricating systems using the function is easier than that using the sigmoid function
Keywords :
backpropagation; cellular neural nets; piecewise-linear techniques; CNN; backpropagation method; cellular neural network; multilayer type neural network; neurons; piecewise-linear function; Backpropagation algorithms; Cellular neural networks; Computer networks; Feeds; Gold; Hopfield neural networks; Multi-layer neural network; Neural networks; Neurons;
Conference_Titel :
Circuits and Systems, 1994. ISCAS '94., 1994 IEEE International Symposium on
Conference_Location :
London
Print_ISBN :
0-7803-1915-X
DOI :
10.1109/ISCAS.1994.409626
