Machine unlearning survey
Yiwen Jiang, Shenglong Liu, Tao Zhao, Wei Li, Xianzhou Gao · Fifth International Conference on Mechatronics and Computer Technology Engineering (MCTE 2022) · 2022
Many online platforms have widely deployed machine learning models as a service. Many of these applications require users to upload their data for model training, but it also induces privacy risks. Once the user wants to leave the application, how to make the application unlearn the uploaded data, which is called machine unlearning, is worthy of study. In this article, we provide a survey of machine unlearning with an approximate and exact guarantee. We summarize the existing machine unlearning approaches and discuss their merits and drawbacks in this field.