Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification

Boyoung Han; Yeonghyeon Kim; Jina Choi; Hojune Shin; Younho Lee

doi:10.1145/3605759.3625262

Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification

Boyoung Han
, Yeonghyeon Kim
, Jina Choi
, Hojune Shin
, Younho Lee

Dept. of Industrial Engineering

Seoul National University of Science and Technology (SNUST)
CryptoLab

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

7 Scopus citations

Abstract

Despite the revolutionary advancement of homomorphic encryption (HE) technology, no efficient fully homomorphic Naive Bayes (NB) classifier that can perform training with HE-encrypted data has been developed without using a decryption function. In this study, we propose an approximate homomorphic logarithm calculation method with a relative error of less than 0.01% on average. Using the SIMD function of the underlying HE scheme, the logarithm values for thousands of encrypted probability values can be calculated in approximately 2.5s with the help of a GPU. Based on this, we propose an efficient fully homomorphic NB method. The proposed NB classifier could complete the training on the breast cancer dataset considered within approximately 14.3s, and perform inference for a query in 0.84s. This is estimated around 28 times faster compared to the recent privacy-preserving NB classifier supporting an analogous level of security by Liu et al. in 2017 on the same computational environment and the same CKKS HE operations performance.

Original language	English
Title of host publication	WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography
Publisher	Association for Computing Machinery, Inc
Pages	91-102
Number of pages	12
ISBN (Electronic)	9798400702556
DOIs	https://doi.org/10.1145/3605759.3625262
State	Published - 26 Nov 2023
Event	11th Annual Workshop on Encrypted Computing and Applied Homomorphic Cryptography, WAHC 2023 - Copenhagen, Denmark Duration: 26 Nov 2023 → …

Publication series

Name	WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography

Conference

Conference	11th Annual Workshop on Encrypted Computing and Applied Homomorphic Cryptography, WAHC 2023
Country/Territory	Denmark
City	Copenhagen
Period	26/11/23 → …

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

ckks
fully homomorphic encryption.
naive bayes classifier
privacy-preserving machine learning

Access to Document

10.1145/3605759.3625262

Cite this

Han, B., Kim, Y., Choi, J., Shin, H., & Lee, Y. (2023). Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification. In WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography (pp. 91-102). (WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography). Association for Computing Machinery, Inc. https://doi.org/10.1145/3605759.3625262

Han, Boyoung ; Kim, Yeonghyeon ; Choi, Jina et al. / Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification. WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography. Association for Computing Machinery, Inc, 2023. pp. 91-102 (WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography).

@inproceedings{33a0bedb49884cf0a4b6705a7c302d00,

title = "Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification",

abstract = "Despite the revolutionary advancement of homomorphic encryption (HE) technology, no efficient fully homomorphic Naive Bayes (NB) classifier that can perform training with HE-encrypted data has been developed without using a decryption function. In this study, we propose an approximate homomorphic logarithm calculation method with a relative error of less than 0.01\% on average. Using the SIMD function of the underlying HE scheme, the logarithm values for thousands of encrypted probability values can be calculated in approximately 2.5s with the help of a GPU. Based on this, we propose an efficient fully homomorphic NB method. The proposed NB classifier could complete the training on the breast cancer dataset considered within approximately 14.3s, and perform inference for a query in 0.84s. This is estimated around 28 times faster compared to the recent privacy-preserving NB classifier supporting an analogous level of security by Liu et al. in 2017 on the same computational environment and the same CKKS HE operations performance.",

keywords = "ckks, fully homomorphic encryption., naive bayes classifier, privacy-preserving machine learning",

author = "Boyoung Han and Yeonghyeon Kim and Jina Choi and Hojune Shin and Younho Lee",

note = "Publisher Copyright: {\textcopyright} 2023 ACM.; 11th Annual Workshop on Encrypted Computing and Applied Homomorphic Cryptography, WAHC 2023 ; Conference date: 26-11-2023",

year = "2023",

month = nov,

day = "26",

doi = "10.1145/3605759.3625262",

language = "English",

series = "WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography",

publisher = "Association for Computing Machinery, Inc",

pages = "91--102",

booktitle = "WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography",

}

Han, B, Kim, Y, Choi, J, Shin, H & Lee, Y 2023, Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification. in WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography. WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography, Association for Computing Machinery, Inc, pp. 91-102, 11th Annual Workshop on Encrypted Computing and Applied Homomorphic Cryptography, WAHC 2023, Copenhagen, Denmark, 26/11/23. https://doi.org/10.1145/3605759.3625262

Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification. / Han, Boyoung; Kim, Yeonghyeon; Choi, Jina et al.
WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography. Association for Computing Machinery, Inc, 2023. p. 91-102 (WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification

AU - Han, Boyoung

AU - Kim, Yeonghyeon

AU - Choi, Jina

AU - Shin, Hojune

AU - Lee, Younho

PY - 2023/11/26

Y1 - 2023/11/26

N2 - Despite the revolutionary advancement of homomorphic encryption (HE) technology, no efficient fully homomorphic Naive Bayes (NB) classifier that can perform training with HE-encrypted data has been developed without using a decryption function. In this study, we propose an approximate homomorphic logarithm calculation method with a relative error of less than 0.01% on average. Using the SIMD function of the underlying HE scheme, the logarithm values for thousands of encrypted probability values can be calculated in approximately 2.5s with the help of a GPU. Based on this, we propose an efficient fully homomorphic NB method. The proposed NB classifier could complete the training on the breast cancer dataset considered within approximately 14.3s, and perform inference for a query in 0.84s. This is estimated around 28 times faster compared to the recent privacy-preserving NB classifier supporting an analogous level of security by Liu et al. in 2017 on the same computational environment and the same CKKS HE operations performance.

AB - Despite the revolutionary advancement of homomorphic encryption (HE) technology, no efficient fully homomorphic Naive Bayes (NB) classifier that can perform training with HE-encrypted data has been developed without using a decryption function. In this study, we propose an approximate homomorphic logarithm calculation method with a relative error of less than 0.01% on average. Using the SIMD function of the underlying HE scheme, the logarithm values for thousands of encrypted probability values can be calculated in approximately 2.5s with the help of a GPU. Based on this, we propose an efficient fully homomorphic NB method. The proposed NB classifier could complete the training on the breast cancer dataset considered within approximately 14.3s, and perform inference for a query in 0.84s. This is estimated around 28 times faster compared to the recent privacy-preserving NB classifier supporting an analogous level of security by Liu et al. in 2017 on the same computational environment and the same CKKS HE operations performance.

KW - ckks

KW - fully homomorphic encryption.

KW - naive bayes classifier

KW - privacy-preserving machine learning

UR - https://www.scopus.com/pages/publications/85180123878

U2 - 10.1145/3605759.3625262

DO - 10.1145/3605759.3625262

M3 - Conference contribution

AN - SCOPUS:85180123878

T3 - WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography

SP - 91

EP - 102

BT - WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography

PB - Association for Computing Machinery, Inc

T2 - 11th Annual Workshop on Encrypted Computing and Applied Homomorphic Cryptography, WAHC 2023

Y2 - 26 November 2023

ER -

Han B, Kim Y, Choi J, Shin H, Lee Y. Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification. In WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography. Association for Computing Machinery, Inc. 2023. p. 91-102. (WAHC 2023 - Proceedings of the 11th Workshop on Encrypted Computing and Applied Homomorphic Cryptography). doi: 10.1145/3605759.3625262

Fully Homomorphic Privacy-Preserving Naive Bayes Machine Learning and Classification

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this