Comparative Study of Clustering Distance Measures to Determine Neural Network Architectures

Abstract

This paper presents a comparative study of clustering distance measures to determine the architecture of a neural network. A number of different Distance Measures have been proposed to measure ‘distance’ for the cluster analysis, which will be applied on the dataset used to train the neural network. The goal of this study is to select the optimal number of clusters which depends on the best choice of clustering distance measure because the results of cluster analysis are used to determine the number of hidden layers and the number of hidden neurons. A particular criterion presented in this paper to link between the number of cluster obtained and the number of hidden layers of a neural network.