Title :
Optimal layering of neurons
Author :
Brandt, Robert D. ; Lin, Feng
Author_Institution :
Intelligent Devices Inc., Glen Ellyn, IL, USA
Abstract :
Given the number of input, output, and hidden neurons in a neural network, what is the optimal layering of hidden neurons that will maximize the topological capacity of the neural network? Here we define the “topological capacity” of a neural network to be the number of connections between neurons. The answer to this question is very interesting. For hierarchical networks, having as many layers as possible is best; while for strictly layered networks, having one or two layers is best
Keywords :
neural nets; hidden neurons; hierarchical networks; input neurons; output neurons; strictly layered networks; topological capacity; Neural networks; Neurons; Upper bound;
Conference_Titel :
Intelligent Control, 1996., Proceedings of the 1996 IEEE International Symposium on
Conference_Location :
Dearborn, MI
Print_ISBN :
0-7803-2978-3
DOI :
10.1109/ISIC.1996.556251
