Efficient Dyslexia Detection Using a Subset of Facial Features: a Comparative Analysis with ANN and CNN

EasyChair Preprint 14536

Abstract

Dyslexia, a learning disability that affects reading and writing skills, has been the subject of extensive research, particularly in the field of machine learning-based classification. In this study, we focused on a subset of features extracted using Google ML Kit, to enhance the efficiency of dyslexia classification. The selected features resulted in about 97% of the total number of features that had been collected. We employ two classification methods, Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN), to evaluate the effectiveness of this feature subset. Preliminary results indicate that the selected features contribute significantly to the classification accuracy, with the CNN model outperforming the ANN in most scenarios. This focused approach not only streamlines the feature selection process but also demonstrates the potential for more targeted and efficient dyslexia detection methods. Our findings suggest that reducing the feature set while maintaining high classification performance is feasible, paving the way for more practical applications in real-world settings.

Keyphrases: Dyslexia Classification, Feature subset selection, Google ML Kit, Supervised Machine Learning, learning disabilities

@booklet{EasyChair:14536,
  author    = {Günet Eroğlu and Mhd Raja Abou Harb},
  title     = {Efficient Dyslexia Detection Using a Subset of Facial Features: a Comparative Analysis with ANN and CNN},
  howpublished = {EasyChair Preprint 14536},
  year      = {EasyChair, 2024}}