The lecturer, Director of the Division of Scientific Affairs and Cultural Relations in the Department of Computer Science, Eng. Taiba Walaa El-Din Khairy Saeed, obtained A global certificate of participation and appreciation through its participation in the research entitled "Improving the problem of software MAX regression and font recognition." Handwritten using the Tensor Flow Library at a global conference in Australia sponsored by IEEE and in collaboration Global Scholarships with Academic Cherub Affiliated to Charles Sturt University (CSU) in Australia on a voluntary basis. The aim of this paper is to improve the classification accuracy of current handwritten number systems, and thus improve their efficiency. Consists The proposed system has a decision function enhanced by adding the "Likelihood Bias" function. The job adds negative weights to categories Outputs that have a high positive bias and add positive weight to classes of outputs that have a high negative bias for Neutralizes the effect of this high negative bias. Therefore, the Bayesian Classifier function has been improved and thus the classification accuracy is improved, thus It will further improve the performance of the multi-class likelihood classification. A 5.6% increase in the overall accuracy of number classification was observed Handwritten using the revised National Institute of Standards and Technology (MNIST) data set.