Least Squares Support Vector Machine Classifier is proposed based on Support VectorMachine Classifier by the inequality constraints to equality constraints. Recently, manystudies show the structural information in data is very important to design a good classifier.The structure information in data has not been exploited in the Least Squares SupportVector Machine Classifier (LSSVM). We focus on the above issue of the LSSVM, structuralleast squares support vector machine (SLSSVM) is proposed. The structure information isconsidered by incorporating the covariance matrix into the objective function.However, the SLSSVM is sensitive to the outliers and in order to reducing sensitive tothe outliers, structured weighted least squares support vector machine(SWLSSVM) isproposed. The structure information is considered by incorporating the covariance matrix intothe objective function, and for reducing sensitive to the outliers, each sample is assigned bythe different weights, which according to the distances of samples to the center of samples.The experimental results show that the SWLSSVM and SLSSVM are more superior to theLSSVM and SVM in generalization performances. |