A New Optimization Method of Large-Scale SVMs Based on Kernel Distance Clustering

Against the low efficiency of training on large-scale SVM, a reduction approach based on kernel distance clustering is proposed. The kernel distance's formulation is brought in to cluster the highly-dimensioned dataset, and the clustering step will reduce a large amount of unsupport vectors during training, thereby, the training time will decrease. The experiments show that this new training algorithm is able to speed up the training process and improve the classification's precision.