https://machinelearningmodels.org/privacy-in-machine-learning-with-adversarial-regularization/