accuracy improvement in differentially private logistic regression: a pre-trainingapproach

Machine learning (ml) models can memorize training datasets. as a result, training ml models on private datasets can lead to the violation of individuals’ privacy. differential privacy (dp) is a rigorous privacy notion to preserve the privacy of the underlying training datasets. however, training ml models in a dp framework usually degrades the accuracy of ml models. this paper aims to increase the accuracy of a dp logistic regression (lr) via a pre-training module. in more detail, we initially pre-train our lr model on a public training dataset without any privacy concern. then, we fine-tune our dp-lr model with the private dataset. in the numerical results, we show that adding a pre-training module significantly improves the accuracy of the dp-lr model.