Abstract:Data mining technology based on encryption often has too high computational complexity.However,as a data perturbation technology,differential privacy can effectively reduce the computational complexity and allow users to analyze the overall value of data.This article briefly analyzes the related technologies of differential privacy in intelligent data mining systems with sensitive information.First,the basic concepts of differential privacy are introduced.Second,differential privacy is introduced for the three major categories of frequent item mining,regression and classification in the application of sensitive data mining,in deep learning.Finally,the performance,advantages and disadvantages of the latest research methods are compared,which provides a reference direction for further research on privacy-protected data mining technology.