There are 5 sample files to be used as training set and testing set. Suppose you want to view results, just click on the above button. "Original" refers to the original dataset, "SVD" refers to the tranformed file through SVD processing, "Sparsified SVD" refers to the transformed file through sparsified SVD, "Uniformlly" refers to the transformed file after adding uniformally distributed noise, and "Normally" refers to the transformed file after adding normally distributed noise. We just gave the final results. If you are interested in the accuracy comparison as well, please contact us.