PERBANDINGAN METODE SOM (SELF ORGANIZING MAP) DENGAN PEMBOBOTAN BERBASIS RBF (RADIAL BASIS FUNCTION)

Abstract

In many clustering systems many methods was used to cluter-ization, one of which is the SOM (Self Organizing Maps). In our study we used two approaches. The first  approach was a lawyer-cluster's using SOM-RBF used in the training data and could be expected to result in better cluster. And the second approach clustering was used of SOM.Comparison of both methods is based on the application of the data derived from  the  dataset  movielens.org  site.  Comparative  assessment  using  three scenarios, namely the MSE as a stop condition on the running time, the MSE as the  stop  condition of the epoch and the learning rate, and MSE as the stop condition of the actual value of the MSE. With this running time is detected which is more rapid approach to the time span for extracting training data. Based on the results of experiments performed using 500 data, which is applied to clusters 3 and 4 lead to the conclusion that the first approach has the value of MSE is actually closer to the absolute value of MSE as compared to the second approach.