Envoyer la notice par courriel: Supporting Regularized Logistic Regression Privately and Efficiently