Recognition of Intrusive Alphabets to the Arabic Language Using a Deep Morphological Gradient

Abstract

The optical character recognition field was one of the key areas of evidence for deep learning methods and has become one of the most successful applications of this technology. Despite that the Arabic is among the most spoken languages in the world today. the optical recognition of Arabic manuscript characters by the algorithms of deep learning remains insufficient. Recently, some studies are moving towards this side and give remarkable results either for the recognition of alphabets or Arabic numbers. We present a deep architecture to solve the problem of the recognition of intrusive handwritten characters to Arabic language. We use a fusion between the morphological gradient method to detect the contours of the alphabets, and multi-layer perceptron (MLP) network with regularization parameters like batch normalization. We apply this model for a database that we created. The classification accuracy was 100% with a very small loss of 0.2%.

Using Natural Language Processing for Programming Language Code Classification with Multinomial Naive Bayes

“…Kiyak et al [21] combined text and image-based comparisons to identify programming languages with high accuracy (over 93.5%) across three datasets. For the Arabic language, research [22] achieved a classification accuracy of 100% using a deep learning model based on multi-layer perceptron neural networks. The concept of graph-based representations for programming language source code was introduced by Allamanis et al [23], highlighting their ability to capture structural and semantic information for classification purposes.…”