Sample-Efficient Private Learning of Mixtures of Gaussians

Neural Information Processing Systems 

We study the problem of learning mixtures of Gaussians with approximate differential privacy.