Pengenalan Pola Tulisan Tangan Suku Kata Aksara Sasak Menggunakan Metode Integral Projection dan Neural Network

Abstract

This paper presents sasak ancient scripts using integral projection and neural network. The purpose of using these two methods is to find out how to work this methods and how much accuracy is obtained in pattern recognition of sasak ancient scripts. The data used is 1260 handwritten image data. Testing is done by knowing the effect of the number of nodes in a hidden layer and the effect of the number of hidden layers in a network. The highest accuracy on average is in the use of 2 hidden layers where 21 nodes for the first hidden layer and 14 nodes for the second hidden layer. The experiment resulted in the obtained accuracy rate of 41.38%.