Implementation of Backpropagation Method for Identified Javanese Scripts

• Published in: Journal of physics. Conference series Vol. 1477; no. 3; pp. 32020 - 32025
• Main Authors: Prihandi, Ifan, Ranggadara, Indra, Dwiasnati, Saruni, Sari, Yunita Sartika, Suhendra
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.03.2020
• Subjects: Algorithms; Artificial neural networks; Back propagation; Back propagation networks; Culture; Machine learning; Neural networks; Physics
• DOI: 10.1088/1742-6596/1477/3/032020
• ISSN: 1742-6588

Literacy is one important component in a culture, one of which is Javanese culture. Javanese script is used as the main script that is used daily by most people in Central Java. Along with the development of human civilization and the expansion of global life, Javanese script began to be abandoned, even by the Javanese themselves. The use of Javanese characters is reduced because of the difficulty of learning Javanese characters themselves. Java character sets consist of basic characters, numbers, complementary characters, and so on. Javanese script is a traditional Indonesian script known as Hanacaraka or Carakan. Backpropagation is a systematic method of artificial neural networks using supervised learning algorithms and is usually used by perceptron with multiple screen layers to change the weights in the hidden layer. Text recognition application in alphabetical letters with neural networks. In the research we made, the training results of this application are good, because it can recognize letters well, the maximum characters that can be read by the application are 128 characters [Latin characters], based on ASCII code pages, this text recognition application can recognize letters and numbers, namely AZ and 0-9 with a percentage of success of up to 90%, Input characters to teach on the form train can only teach 1 character only.