Identification of Different Arabic Dialects Using Randomly Multimodal Deep Learning (RMDL) Approach on AOC Database

Like other languages, the Arabic dialect has both spoken and written forms. The first form is the true mother tongue of Arabic speakers explored on a daily basis. As for the second form, it represents the dialect of Modern Standard Arabic (MSA) which mainly constitutes the content of most Arabic databases due to its predominance in written form. However, the large size of Arabic text databases requires robust and accurate machine learning methods. Random Multimodal Deep Learning (R MDL) is one such deep learning approach that is explored in this article to address the issue of finding the best deep learning architecture and structure while simultaneously maintaining improvement accuracy and robustness for the classification and identification of Arabic dialects (MSA, Gulf (GLF), Egypt (EGY) and Levantine (Lev)) via a set of deep learning architectures. Experimental tests were carried out on the Arabic Online Commentary (AOC) database and showed good results in terms of performance evaluated using the RMDL approach compared to those obtained with other deep learning algorithms.