Deep Learning Autoencoder Approach for Handwritten Arabic Digits Recognition

TL;DR

This paper introduces a stacked autoencoder-based unsupervised learning method for recognizing handwritten Arabic digits, achieving high accuracy and addressing challenges like handwriting variation.

Contribution

The work presents a novel application of stacked autoencoders for Arabic digit recognition, demonstrating significant accuracy improvements on a large public database.

Findings

Achieved an average accuracy of 98.5% on MADBase database.

SAE improves performance across various classifiers.

Addresses challenges of handwriting variation and large datasets.

Abstract

This paper presents a new unsupervised learning approach with stacked autoencoder (SAE) for Arabic handwritten digits categorization. Recently, Arabic handwritten digits recognition has been an important area due to its applications in several fields. This work is focusing on the recognition part of handwritten Arabic digits recognition that face several challenges, including the unlimited variation in human handwriting and the large public databases. Arabic digits contains ten numbers that were descended from the Indian digits system. Stacked autoencoder (SAE) tested and trained the MADBase database (Arabic handwritten digits images) that contain 10000 testing images and 60000 training images. We show that the use of SAE leads to significant improvements across different machine-learning classification algorithms. SAE is giving an average accuracy of 98.5%.