Comparison of Support Vector Machines and Multilayer Perceptrons in the Classification Process: A Case Study of Heart Disease Analysis

Abstract

Pattern Recognition is an important area in computer science that maps data to predefined concepts. Support Vector Machines (SVM) are particularly effective due to their ability to identify the optimal hyperplane that separates two classes in the feature space. Unlike neural networks, which look for a separating hyperplane, SVM determines the best hyperplane in the input space. SVM primarily serves as a linear classifier but can also address non-linear problems through the kernel trick, enabling high-dimensional operations. This paper delves into the foundational principles of SVM and its applications, specifically in classifying heart disease symptoms in individuals. The research includes the implementation of Gaussian Radial Basis Function (RBF) and Polynomial (POLY) kernel functions, along with various parameters affecting SVM performance. Additionally, a comparative analysis with Multilayer Perceptron (MLP) for data classification is presented to evaluate the effectiveness of the proposed kernel functions.