A New Data Classification Approach Based on Deep Learning Techniques

Abstract

ABSTRACT In this study, we developed a new model based on Convolutional Neural Network (CNN) and the power of signal function for Covid-19 detection. The presented method consists of three parts: CNN, power of the signal, and classifiers. The aim of applying CNN is to extract high-level and sensitive features from input x-ray images. The CNN pre-trained model Alexnet is used in this section as a features extractor and the extracted features are wired to the power of signal function that is used to calculate the power of each period and reduce the size of input features. Then, the extracted features by the power of the signal are wired directly to the classifiers. The obtained results have 95.5% accuracy which is remarkable when compared with several states of art studies presented in this field.