Weighted least square support vector machine integrated with parameter optimization and its application in chemical process

Considering that outliers can disrupt the correlation structure of least square support vector machine (LS-SVM), and that the parameters of LS-SVM play an important role in the performance, a novel weighted least square support vector machine integrated with parameter optimization is proposed to obtain the optimal parameters and to eliminate the effect of outliers. Several LS-SVM variants are applied in simulation experimentation and chemical process respectively to demonstrate the satisfactory performance of the proposed method.