Rademacher Observations, Private Data, and Boosting
Published on Dec 05, 20151524 Views
The minimization of the logistic loss is a popular approach to batch supervised learning. Our paper starts from the surprising observation that, when fitting linear classifiers, the minimization of th