BCP and ZQP Strategies to Reduce the SVM Training Time

The Support Vector Machine (SVM) is awell known method used for classification, regression and density estimation. Training a SVM consists in solving a Quadratic Programming (QP) problem. The QP problemis very resource consuming (both computational time and computational memory), because the quadratic form is dense and the memory requirements grow square the number ofdata points.In order to increase the training speed of SVM's, this paperproposes a combination of two methods, the BCP algorithm(Barycentric Correction Procedure), [15], to find, heuristically,training points with a high probability to be Support Vectors,and the ZQP algorithm, [10], to solve the reduced problem.