Implementation of One Orde Extraction for Identification of Coal Batik Method with Backpropagation Method

Abstract

The coastal areas of Java Island covering the cities of Brebes, Cirebon, Pekalongan, Lasem and Madura have various patterns of batik motifs. Based on the pattern of coastal batik motifs can be distinguished into batik geometric a non-geometry. Classification of coastal batik motif using Backpropagation algorithm by determining the value of learning rate and momentum during training data. Input data used in the form of statistical characteristics obtained from the formation of GLCM values. Statistical characteristics used include mean, standard deviation, curtosis, skewness and entropy. While the best learning rate is obtained on the number 0.5 and the momentum 1.0 on the geometry of batik motif. While the best non-geometric motif of learning rate is obtained on the number 0, 5 and momentum 1.0. The number of neurons used in the training of both motives affects the epoch value (the number of iterations) and the resulting error.