Proses Pengelompompokan Saraf Menggunakan Jaringan Saraf Tiruan (JST) dengan Algoritme Self-Organizing Maps (SOM)

Abstract

Artificial Neural Network Method (ANN) with Self-Organizing Maps (SOM) algorithm is one of the methods using unsupervised learning method. This means that this method does not have a variable Y or without any sample data output. So the SOM algorithm can group the input into output in the form of cluster. The validity index used is the Davies-Bouldin Index (DBI). As an application material, data used PMB UNISBA 2016 obtained from Data Processing Administration UNISBA. The result of clustering shows that clustering of data having the smallest DBI of 1,4463 is the output 3 or group size formed by 3 clusters. The results obtained through the initialization of learning rate ( ) of 0,9 and decreased learning rate ( of 0,04 and iteration as much as 500 times. From the clustering results obtained the average of the centroid of each cluster is 86,9009 ( cluster to 1), 83,3775 ( cluster to 2), and 80,0384 ( cluster to 3).