Methods of Decreasing the Number of Support Vectors via k-Mean Clustering