Empirical Determination of Sample Sizes for Multi-layer Perceptrons by Simple RBF Networks

It’s well known that the computing time to train multilayer perceptrons is very long because of weight space of the neural networks and small amount of adjustment of the wiights for convergence. The matter becomes worse when the size of training data set is large, which is common in data mining tasks. Moreover, depending on samples, the performance of neural networks change. So, in order to determine appropriate sample sizes for multilayer perceptrons this paper suggests an effective approach with the help of simple radial basis function networks that work as a guide. Experiments with the two different data sets that may represent business and scientific domain well showed the effectiveness of the suggested method. Key-Words: multilayer perceptron, sample size, radial basis function network, data mining