Subsequently, we need to keep the dimension of the populationand halve the concatenated matrix to obtain. To do so, we first introduce the accuracy and the fitness function of the classification process as equation (25).Owing to the feature selection is considered as a multi-objective problem, it is necessary to minimize the number of features and maximize the classification accuracy of a given classifier at the same time. Therefore, equation (24) is used to evaluate the joint feature matrix H, expressed as follows: