Vol.1, No.3, 2019, pp.145-150, doi:10.32604/jbd.2019.08706
OPEN ACCESS
ARTICLE
A Privacy Preserving Deep Linear Regression Scheme Based on Homomorphic Encryption
  • Danping Dong1, *, Yue Wu1, Lizhi Xiong1, Zhihua Xia1
1 School of Computer and Software, Nanjing University of Information Science and Technology, Nanjing, 210044, China.
* Corresponding Author: Danping Dong. Email: dongdp@139.com.
Abstract
This paper proposes a strategy for machine learning in the ciphertext domain. The data to be trained in the linear regression equation is encrypted by SHE homomorphic encryption, and then trained in the ciphertext domain. At the same time, it is guaranteed that the error of the training results between the ciphertext domain and the plaintext domain is in a controllable range. After the training, the ciphertext can be decrypted and restored to the original plaintext training data.
Keywords
Linear regression, somewhat homomorphic encryption, machine learning.
Cite This Article
Dong, D., Wu, Y., Xiong, L., Xia, Z. (2019). A Privacy Preserving Deep Linear Regression Scheme Based on Homomorphic Encryption. Journal on Big Data, 1(3), 145–150.