Naive Bayes Classification based on Differential Privacy
Wenru Tang, Yihui Zhou, Zhenqiang Wu, Laifeng Lu, Mingshuang Li · 2019
Data mining has a wide range of applications in the real world. However, it is possible to disclose the private information of users in the process of data mining. Therefore, it is of great significance to protect the users' privacy while mining the knowledge behind the data. In this paper, we propose a Naive Bayes classification method based on differential privacy. For nominal attributes, we add Laplace noise to the count. For numerical attributes, we add Laplace noise to the mean, standard deviation, and scale parameter, and then use the noisy parameters to calculate the prior probability and conditional probability. For numerical attributes, we assume that they follow Gaussian, Laplace, or lognormal distribution, and apply our algorithms to compare utilities.