A BA-based algorithm for parameter optimization of Support Vector Machine

• Published in: Pattern recognition letters Vol. 93; pp. 13 - 22
• Main Authors: Tharwat, Alaa, Hassanien, Aboul Ella, Elnaghi, Basem E.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01-07-2017; Elsevier Science Ltd
• Subjects: Algorithms; Artificial intelligence; Bat algorithm (BA); Classification; Genetic algorithms; Learning algorithms; Machine learning; Mathematical models; Model accuracy; Optimization algorithms; Parameter optimization; Parameters; Particle swarm optimization; Regression analysis; Support Vector Machine (SVM); Support vector machines; Swarm intelligent; Swarm intelligent; Optimization algorithms; Bat algorithm (BA); Parameter optimization; Support Vector Machine (SVM); Classification
• ISSN: 0167-8655; 1872-7344
• DOI: 10.1016/j.patrec.2016.10.007

•SVM in the framework of classification was highlighted.•Bat Algorithm (BA) was proposed to search for SVM parameters.•The proposed algorithm was applied on nine standard classification datasets.•The results demonstrated lower classification error rates.
Support Vector Machine (SVM) parameters such as kernel parameter and penalty parameter (C) have a great impact on the complexity and accuracy of predicting model. In this paper, Bat algorithm (BA) has been proposed to optimize the parameters of SVM, so that the classification error can be reduced. To evaluate the proposed model (BA-SVM), the experiment adopted nine standard datasets which are obtained from UCI machine learning data repository. For verification, the results of the BA-SVM algorithm are compared with grid search, which is a conventional method of searching parameter values, and two well-known optimization algorithms: Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). The experimental results proved that the proposed model is capable to find the optimal values of the SVM parameters and avoids the local optima problem. The results also demonstrated lower classification error rates compared with PSO and GA algorithms.