Journal ID : TRKU-08-07-2020-10867
[This article belongs to Volume - 62, Issue - 06]
Total View : 325

Title : Development Optimization Parameter Selection in the Least Square Support Vector Machine Classification Algorithm (LS-SVM)

Abstract :

Support vector machine (SVM) is one of the flagship methods in machine learning because SVM have good results in terms of classification and prediction. Classification Analysis is the process of finding the best model of the classifier to predict classes of an object or data whose class label is unknown. The study aims to select accurate classification methods in classification. The study used the Least Support Vector Machine (LS-SVM) method. LS-SVM is better compared to standard SVM in computation processes, rapid convergence, and high precision. On LS-SVM it has several parameters. Optimal parameter delivery on the LS-SVM method can affect increased classification accuracy. As for the various kinds of giving optimal parameter selection methods on LS-SMVS such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Gravitational Search Algorithm (GSA), Gird Search, and Particle Swarm Optimization and Gravitational Seach Algorithm (PAO-GSA). The study results show that the selection of more optimal parameters can result in significantly increased classification and prediction accuracy and more even computational process times. PSO-GSA is a method of selecting optimization parameters that can be used to improve classification accuracy

Full article

//