Abstract:  In statistical learning theory, generalization error is used to quantify the degree to which a supervised machine learning algorithm may overfit to training data.