IDENTIFIKASI GORESAN DASAR MANDARIN DENGAN METODE MULTILAYER PERCEPTRON

• Main Author: Dewi
• Format: Article
• Language: Indonesian
• Published: LPPM Universitas Potensi Utama 2015-02-01
• Series: CSRID Journal
• Subjects: Basic Stroke; Mandarin Language
• Online Access: http://csrid.potensi-utama.ac.id/ojs/index.php/CSRID/article/view/61

Mandarin Language is the second international language after English Language. Mandarin Language is different with English Language. Mandarin Language consists of stroke, intonation and pin yin. The basic strokes in Mandarin Language are eleven strokes. In this research, author identifies the basic stroke of Mandarin using Multilayer Perceptron to determine how the accuracy of Multilayer Perceptron to recognize the strokes. Data of the basic stroke of Mandarin that used are strokes from several different people.The data has been saved in image with size 80x80 pixel and changed into black and white image. Then taking the FFT and Mean Citra value from the image. The next step is training the data, determining the target and implementation the multilayer perceptron method. The accuracy that reached by multilayer perceptron method in identifying the basic stroke of Mandarin is 59.09% with 45 node of hidden layer. The node amount of hidden layer very affect the output value.