Techniques that protect user data privacy during the machine learning process, without compromising the utility of the models.
Generality: 760
