Differential Privacy Protection Technology Review for Machine Learning

With the development of information and communication technology, machine learning has become an indispensable technical tool in many industrial and research fields. However, there is a large amount of personal information in the data required for machine learning, bringing risks and challenges into privacy protection. Differential privacy protection is a perturbation based privacy protection mechanism widely used in machine learning. In this paper, the common privacy protection technologies based on encryption and perturbation in machine learning are firstly summarized. Secondly, the supervised and unsupervised differential privacy protection technologies for machine learning are distinguished. Finally, the future development directions of this field are offered.