Sample Complexity Bounds on Differentially Private Learning via Communication Complexity

In this work we analyze the sample complexity of classification by differentially private algorithms. Differential privacy is a strong and well-studied notion of privacy introduced by Dwork et al. (20