Hybrid Neutrosophic Deep Learning Model for Enhanced Arabic Handwriting Recognition

Abstract

 Recognizing handwritten Arabic characters poses a significant challenge due to the 
complexities of the cursive script and the visual similarities between characters. While deep learning 
techniques have shown substantial promise, advancements in model architectures are essential to 
further enhance performance. Neutrosophic Sets (NS) have demonstrated their potential in 
improving classification models by effectively handling indeterminate and inconsistent data. This 
paper introduces a novel approach that integrates Neutrosophic Sets with a hybrid deep learning 
model, combining Convolutional Neural Networks (CNNs) with Bidirectional Recurrent Neural 
Networks (Bi-LSTM and Bi-GRU). This integration allows for the extraction of spatial features and 
modeling of temporal dynamics in handwritten Arabic text. Experiments conducted on the Hijjaa 
and AHCD datasets revealed that the NS_CNN_Bi-LSTM model achieved an accuracy of 92.38% on 
the Hijjaa dataset, while the NS_CNN_Bi-GRU model attained 97.38% accuracy on the AHCD 
dataset, outperforming previous deep learning approaches. These results highlight the significant 
performance improvements achieved through advanced temporal modeling and contextual 
representation, without the need for explicit segmentation. The findings contribute to the ongoing 
development of highly accurate and sophisticated deep learning systems for Arabic handwriting 
recognition, with broad applications in areas requiring efficient extraction of text from handwritten 
documents.

DOI: 10.5281/zenodo.13763213