Pengaruh Variasi Jumlah Data Pelatihan SVM Terhadap Unjukkerja pada Sistem Pengenalan Pola Tulisan Tangan Aksara Jawa Nglegeno

Abstract

Support Vector Machines (SVM) is one of many classification methods implemented on pattern recognition, including handwritten character recognition system. Numbers of traning set will affect the performance of the recognition system. A basic Jawanessehandwritten character recognition system based on SVM used to show the effect of training set numbers variation on system performance. The experiment result indicating that SVM has good generalization on limiting training set (on a data set, 30% data used for training set can give more than 90% success rate on all data set).