Research Article
Non-interactive Privacy-Preserving Naïve Bayes Classifier Using Homomorphic Encryption
@INPROCEEDINGS{10.1007/978-3-030-96791-8_14, author={Jingwei Chen and Yong Feng and Yang Liu and Wenyuan Wu and Guanci Yang}, title={Non-interactive Privacy-Preserving Na\~{n}ve Bayes Classifier Using Homomorphic Encryption}, proceedings={Security and Privacy in New Computing Environments. 4th EAI International Conference, SPNCE 2021, Virtual Event, December 10-11, 2021, Proceedings}, proceedings_a={SPNCE}, year={2022}, month={3}, keywords={Privacy-preserving data mining Homomorphic encryption Na\~{n}ve Bayes classifier}, doi={10.1007/978-3-030-96791-8_14} }- Jingwei Chen
Yong Feng
Yang Liu
Wenyuan Wu
Guanci Yang
Year: 2022
Non-interactive Privacy-Preserving Naïve Bayes Classifier Using Homomorphic Encryption
SPNCE
Springer
DOI: 10.1007/978-3-030-96791-8_14
Abstract
In this paper, we propose a privacy-preserving naive Bayes classifier based on a leveled homomorphic encryption scheme due to Brakerski-Gentry-Vaikuntanuthan (BGV). The classifier runs on a server that is also the owner of the model, with input as BGV encrypted data from a client. The classifier produces encrypted classification results which can only be decrypted by the client, whereas the model is only accessible to the server itself. This ensures that the classifier does not leak any private information on either the model of the server or the data and results of the client. More importantly, the classifier does not require any interaction between the server and the client during the classification phase. The main technical ingredient is an algorithm to compute the index of the maximum of an encrypted array homomorphically, which does not require any interaction. The proposed classifier is implemented using a homomorphic encryption library HElib. Preliminary experiments demonstrate the efficiency and accuracy of the proposed privacy-preserving naive Bayes classifier.
