A GA-based feature subset selection and parameter optimization of support vector machine for content - based image retrieval

Abstract

This paper presents the effectiveness of applying genetic algorithm (GA)-based feature subset selection and parameter optimization of support vector machine (SVM) for content-based image retrieval (CBIR). SVM, one of the new techniques for pattern classification, has been widely used in many application areas. The kernel parameters setting for SVM in the training process impacts on the classification accuracy. Feature subset selection is another factor that impacts classification accuracy. The objective of this study is to simultaneously optimize the parameters and feature subset without degrading the SVM classification accuracy using the GA-based approach for CBIR. In this study, we show that the proposed GA-based approach outperforms SVM to the problem of the image classification problem in CBIR. Compared with NN and SVM algorithm, the proposed GA-based approach significantly improves the classification accuracy and has fewer input features for SVM. © Springer-Verlag Berlin Heidelberg 2007.