We split the training data into two disjoint subsets. One of these subsets is used to learn the  parameters. The other subset is our...