Application of Backpropagation Algorithm for Handwriting Recognition. Issue 1 (February 2021)

• Record Type: Journal Article
• Authors: Farlinda, Sustin; Yunus, Muhammad; Pratama, Mudafiq Riyan; Prakoso, Bakhtiyar Hadi; Rahagiyanto, Angga; Susilowati, Dyah; Dasriani, Ni Gusti Ayu; Hakim, Lukmanul; Anam, Khairil
• Is Part Of: Journal of physics. Volume 1783:Issue 1(2021)
• Publication Date: 2021-02
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1783/1/012036
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: Handwritten letter recognition is one form of pattern recognition. The introduction of letters looks simple to humans, but it becomes a tough task for computer programs to complete. In recognizing someone's handwriting, a computer program trained first. This study discusses how a computer recognizes a digital image pattern in the form of handwritten letters using the backpropagation algorithm. The backpropagation algorithm used in making this system is backpropagation with momentum. The network architecture used consists of an input screen with 64 neurons, a hidden screen consisting of 46 neurons, and an output screen with five neurons. Before the recognition process, input image with image format (jpg) processed first, which includes scaling, grayscale, and binarization. The trial results show that the system can recognize handwriting with an accuracy of up to 65%.