Toward an Intelligent Hybrid System Based on Data Analysis and Preprocessing Method

Abstract

In recent years, artificial intelligence (AI) represents a crucial domain or technology that can be found everywhere, it can solve many problems facing the researchers. Hence, the powerfulness of Artificial intelligence contributes enormously to the sustainable growth of various domains (e.g., medical, agriculture, data analysis, and so forth). For that, in this paper, we propose the artificial neural network (ANN), and in particular, the paradigm of the Counter propagation artificial neural network. Our objective is to improve the standard Counter propagation artificial neural network in terms of results and classification accuracy by using the principal component analysis (PCA) for making a modified Counter propagation network through the hybridization of PCA and CPN. So, the PCA is a method among data analysis methods. It can reduce the dimensionality space of original data in small data that contained the new elements or objects, and also with this method we can eliminate the dependence and the obstacles between inputs which allowed to obtain the good classification, this hybridization is used in the data analysis field. Thus, the result shows that the proposed approach of modified CPN gave the best performance than the standard CPN in terms of iteration number, mean error, and classification accuracy.