Weighted least squares twin support vector machine for regression with noise

Twin support vector regression and its extensions have been widely applied in machine learning and data mining. However, most of them can not achieve the satisfactory performances when the noise is involved. To this end, this paper presents a weighted least squares twin support vector regression (WLSTSVR) which can reduce the influence of the noise on prediction accuracy by using the information of the responses of the samples. Furthermore, both offline and online learning algorithms are developed. The experimental results on artificial and benchmark datasets with noise indicate that the both online and offline learning WLSTSVR achieve better prediction accuracy and agreement between estimations and real-values compared with least squares twin support vector regression.