Removing Disparate Impact of Differentially Private Stochastic Gradient Descent on Model Accuracy

Open in new window