Impact of Various Kernels on Support Vector Machine Classification Performance for Treating Wart Disease

Abstract

This study displays the impacts of different types of Kernel functions for improving the learning capacity of Support Vector Machine (SVM) in treating two common types of warts (plantar and common warts). The impacts of four Kernel functions of (SVM): Normalized Polynomial Kernel (NP), Polynomial Kernel (PK), Radial Basis Function Kernel (RBF), and Pearson VII function based Universal Kernel (PUK) have been examined On two sets of data called Cryotherapy and Immunotherapy. Which are universally regarded as the best two methods to treat wart disease using Weka workbench. The first dataset called Cryotherapy consists of information about 90 patients and contains 7 features. The second dataset called Immunotherapy consists of information about 90 patients and contains 8 features. For presenting classification performance impacts each of Accuracy, precision, sensitivity, F-measure and confusion matrix for each kernel has been utilized. According to the results obtained, it was found that each of PUK and RBF performs best classification performance on Cryotherapy dataset with 97.77% accuracy whereas each of PK and PUK performs best classification performance on Immunotherapy dataset with 81.11% accuracy.