SVM parameter selection based on harmony search with an application to hyperspectral image classification

Abstract

Support vector machines is a very efficient and frequently used method in classification of hyperspectral images since they provide high classification accuracy even with a limited number of training samples. The accuracy of SVM depends on choice of kernel parameters. In order to obtain a high classification accuracy, it is vital to optimally determine the kernel parameters. In this work, harmony search method, that has been recently introduced as a heuristic method, will be used to optimally determine the kernel parameters of SVM's radial basis kernel function, and the proposed approach will firstly be experimented on hyperspectral datasets. The proposed approach will be compared to classical grid search strategy and genetic algorithm in terms of computational time and classification accuracy.