Implementing Parallel SMO to Train SVM on CUDA-Enabled Systems