Ensemble learning of Ada boosting Based on Deep Weighting for Classification of Hand written Numbers in Persian (With the doctors' prescription approach)

Converting handwritten data to electronic data is one of the challenges that have been raised over the past years. considering that these data are used in various sciences, solving this challenge is of great importance. one of these sciences is medical science that doctors use in prescriptions. this project tries to classify handwritten numbers with the approach of solving the challenges of handwritten data. over the past years, a variety of solutions have been developed to transform handwritten data based on machine learning. each method categorizes or clusters the data based on the type of data and its use. in this paper, a new approach based on hybrid methods and deep learning is presented for the classification of persian handwritten data. by combining ada and convolution, a deeper examination of the data is performed in basic learning. the purpose of this research is to provide a new technique for classifying images of persian handwritten numbers. the structure of this technique is based on ada boosting, which in turn is based on weak learning. this technique improves learning by repeating weak learning processes and updating weights. meanwhile, the proposed method tried to employ stronger language learners and provide a stronger algorithm by combining these strong learners. this method was evaluated on the hoda standard dataset containing 60,000 training data. the results show that the proposed method has more than 1% less error than the previous methods. in the future, as the base learner develops, new mechanisms can be introduced to improve results with new types of learning.