Handwritten javanese character recognition using descriminative deep learning technique

Research on handwriting recognition using deep learning method has been widely explore by many researchers in the field of computer vision and machine learning. Many researchers mentioned that handwriting recognition using deep learning technique has lead to achieve higher accuracy compared to conventional machine learning techniques. Handwriting character recognition using deep learning has been impalement in Latin, Chinese, Arabic, Persian, and Bangla Character. As for the object of Javanese character is still not much encroached. Since the Javanese Classical Manuscripts contain a variety of scientific treasures that can be taken up in order to be preserved as a valuable heritage possessed from Indonesia. Therefore, in this study, the Javanese character Recognition is applied using Convolutional Neural Network (CNN). CNN is one type of discriminative deep-learning model that is widely used for classification based on supervised learning. CNN method is a very powerful deep learning technique in completing its task to perform data classification with image dataset as an input, because it utilizes pixel neighbor information in feature extraction process with convolution and pooling operation between inputs and kernel. The data than classify using softmax to determine its class based on its features. From the experimental results obtained that the discriminative model of deep learning has confirmed to recognize 20 basic Javanese character with the accuracy 94.57 %.