Support vector machine parameter tuning using firefly algorithm
Eva Tuba, Lazar Mrkela, Milan Tuba · 2016
Performance of the support vector machine strongly depends on parameters settings. One of the most common algorithms for parameter tuning is grid search, combined with cross validation. This algorithm is often time consuming and inaccurate. In this paper we propose the use of stochastic metaheuristic algorithm, firefly algorithm, for effective support vector machine parameter tuning. The experimental results on 13 standard benchmark datasets show that our proposed method achieve better results compared to other state-of-the-art algorithms from literature.