Piecewise multi-classification support vector machines
Olutayo O. Oladunni, Gaurav Singhal · 2009
This paper presents a linear programming formulation for linear and nonlinear piecewise multi-classification support vector machines model for multi-category discrimination of sets or objects. The proposed model can be used to generate linear and nonlinear piecewise classifiers depending on the kernel function employed. Advantages of the linear programming multi-classification SVM formulation include its ability to express a multi-class problem as a single optimization problem and its computational tractability in providing the optimal classification weights for multi-categorical separation. Computational results are provided for validation of the proposed piecewise multi-classification SVM model using four benchmark data sets (GPA, IRIS, WINE, and GLASS data).