Implementation of Convolutional Neural Network – Extreme Learning Machine for Handwriting Recognition of Sundanese Script

Sundanese script is one of the cultural heritages in West Java. Apart from Sundanese which is the official regional language in West Java, Sundanese also has a special script. However, currently very few children understand or can write Sundanese script so it needs to be protected and preserved. In this research, the CNN-ELM method is proposed for recognizing Sundanese script. The CNN-ELM method is a model that utilizes the Convolutional Neural Network (CNN) method to extract image features and the Extreme Learning Machine (ELM) for image classification. This method performs well in terms of training speed and accuracy, and can also work with smaller data sets. The aim of this research is to measure accuracy in recognizing Sundanese script. After testing Sundanese script handwriting several times using CNN-ELM, the best results were obtained with an accuracy of 88% with a total of 30,000 hidden neurons.