Machine Learning's Dropout Training is Distributionally Robust Optimal

Open in new window