The massive data classifiers based on reduced set vectors method

A support vector machine (SVM) is a universal learning machine whose decision surface is parameterized by a set of support vectors. SVMs find application in pattern recognition, regression estimation, and operator inversion for ill-posed problems. SVMs are currently considerably slower in the test phase than other approaches with similar generalization performance. To this important problem, we propose the reduced vector set method to significantly decrease the complexity of the decision rules obtained by the SVM. The proposed method made the SVM test speed competitive with that of other methods. Simulation results for massive and complex data turns out to support our idea.