Adaptive Statistical Learning with Bayesian Differential Privacy

Jun Zhao

In statistical learning, a dataset is often partitioned into two parts: the training set and the holdout (i.e., testing) set. For instance, the training set is used to learn a pred…