Comparative Analysis on Effect of Different SVM Kernel Functions for Classification

The chapter delves into the application of Support Vector Machines (SVM) for classification tasks, focusing on the impact of different kernel functions. It begins by introducing SVM and its core principles, such as the concept of the separating hyperplane and the maximum margin classifier. The paper then explores five types of kernels—linear, Gaussian Radial Basis Function (RBF), polynomial, Laplacian, and sigmoid—and their performance on four distinct datasets. Through extensive experimentation, the study compares the prediction accuracy of each kernel, highlighting the strengths and limitations of each. Notably, the Laplacian kernel emerges as the top performer across all datasets, followed closely by the Gaussian RBF kernel. The chapter also investigates the influence of kernel parameters on prediction accuracy, providing valuable insights into optimizing SVM models. The comparative analysis is supported by detailed experimental work and visualizations, making it a crucial resource for professionals seeking to enhance their understanding of SVM kernel functions and their practical applications.