Optimizing Training Data and Hyperparameters of Support Vector Machines Using a Memetic Algorithm

Abstract

Support vector machine (SVM) is a well-known machine learning algorithm widely used for classification and regression problems. Despite the high prediction rate of this technique in a wide range of real applications, the efficiency of SVM and its classification performance highly depends on the hyperparameters setting as well as the selection of feature subset. Moreover, high memory and computational complexity of SVM training can be a limiting factor for its application on huge dataset. In this work we propose a novel memetic algorithm for support vector machine called SE-SVM to address mentioned problems. We use evolutionary techniques that optimize hyperparameters and select features and training set simultaneously. The algorithm is applied to seven datasets. All of that datasets are binary classification problem. We compare the SE-SVM to different evolutionary algorithms, random search techniques and other well-established classifiers. The experimental results show that the end result obtained by SE-SVM achieves high classification performance with a shorter training and classification time.