Exponentially Biased Discriminant Analysis Based Classification of Covid 19 Chest Images Using Generalized Regression Neural Network

Classification is always an interesting problem in the field of computer vision. In a two class problem, there will be an uncertainty in the classification of adjacent images of two classes. To avoid this uncertainty, an exponentially biased discriminant analysis is proposed for the classification. Initially, the entire database is projected to an exponentially biased space. In this space the data is more separated than the original space. Discriminant analysis is then used to classify the objects in this new space. After the training, the test data are approximated to this space using Generalized Regression Neural Network. The proposed algorithm is evaluated using the database of Covid 19 chest images. A better accuracy is observed for the proposed method by comparing with the normal discriminant analysis. But, this accuracy may not be a very good value. Better scientific approaches on the selection of the exponential biasing may give better classification accuracy.