A study of the use of self-organising map for splitting training and validation sets for backpropagation neural network

Abstract

Validation has been used for the estimation of generalisation error of the backpropagation networks. The simplest way is to divide the available data into training and validation data sets. An approach using the self-organising map is proposed for the selection of the training and validation data sets. The results obtained from this study has shown that the proposed method provides a quick and reliable selection criteria and the overall training time is also reduced by applying the split-sample early stopping approach.