Principal Component Analysis over the Boolean Circuit Within TFHE Scheme

Hyun Jung Doh; Joon Soo Yoo; Mi Yeon Hong; Kang Hoon Lee; Tae Min Ahn; Ji Won Yoon

doi:10.1007/978-981-99-8024-6_3

Principal Component Analysis over the Boolean Circuit Within TFHE Scheme

Hyun Jung Doh, Joon Soo Yoo, Mi Yeon Hong, Kang Hoon Lee, Tae Min Ahn, Ji Won Yoon^*

^*Corresponding author for this work

School of Cybersecurity

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

Abstract

In today’s information-driven world, the need to protect personal data while maintaining efficient data processing capabilities is crucial. Homomorphic Encryption (HE) has emerged as a potential solution, allowing secure processing of encrypted information without compromising privacy. However, current HE schemes suffer from slow processing speeds, especially when dealing with high-dimensional data. This paper focuses on leveraging the PCA technique within the Fast Fully Homomorphic Encryption over the Torus (TFHE) scheme to optimize the speed of subsequent algorithms. TFHE offers the advantage of enabling the homomorphic implementation of any circuit but suffers from extensive execution time. We present tailored PCA algorithms for TFHE, utilizing the power method and eigen-shift techniques to extract eigenvalues and eigenvectors. These techniques provide efficient solutions for performing PCA computations within the TFHE framework. By designing a dedicated PCA circuit using TFHE’s fundamental homomorphic gates, we achieve efficient evaluation times for PCA. The performance analysis shows execution times of 3.42 h for a 16-bit dataset and 12.22 h for a 32-bit dataset, with potential for further improvement.

Original language	English
Title of host publication	Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers
Editors	Howon Kim, Jonghee Youn
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	27-39
Number of pages	13
ISBN (Print)	9789819980239
DOIs	https://doi.org/10.1007/978-981-99-8024-6_3
Publication status	Published - 2024
Event	24th International Conference on Information Security Applications, WISA 2023 - Jeju Island, Korea, Republic of Duration: 2023 Aug 23 → 2023 Aug 25

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14402 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	24th International Conference on Information Security Applications, WISA 2023
Country/Territory	Korea, Republic of
City	Jeju Island
Period	23/8/23 → 23/8/25

Bibliographical note

Publisher Copyright:
© 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Keywords

Homomorphic Encryption
Principal Component Analysis
TFHE

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-981-99-8024-6_3

Cite this

Doh, H. J., Yoo, J. S., Hong, M. Y., Lee, K. H., Ahn, T. M., & Yoon, J. W. (2024). Principal Component Analysis over the Boolean Circuit Within TFHE Scheme. In H. Kim, & J. Youn (Eds.), Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers (pp. 27-39). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14402 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-99-8024-6_3

Principal Component Analysis over the Boolean Circuit Within TFHE Scheme. / Doh, Hyun Jung; Yoo, Joon Soo; Hong, Mi Yeon et al.
Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers. ed. / Howon Kim; Jonghee Youn. Springer Science and Business Media Deutschland GmbH, 2024. p. 27-39 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14402 LNCS).

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

Doh, HJ, Yoo, JS, Hong, MY, Lee, KH, Ahn, TM & Yoon, JW 2024, Principal Component Analysis over the Boolean Circuit Within TFHE Scheme. in H Kim & J Youn (eds), Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14402 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 27-39, 24th International Conference on Information Security Applications, WISA 2023, Jeju Island, Korea, Republic of, 23/8/23. https://doi.org/10.1007/978-981-99-8024-6_3

Doh HJ, Yoo JS, Hong MY, Lee KH, Ahn TM, Yoon JW. Principal Component Analysis over the Boolean Circuit Within TFHE Scheme. In Kim H, Youn J, editors, Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2024. p. 27-39. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-981-99-8024-6_3

Doh, Hyun Jung ; Yoo, Joon Soo ; Hong, Mi Yeon et al. / Principal Component Analysis over the Boolean Circuit Within TFHE Scheme. Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers. editor / Howon Kim ; Jonghee Youn. Springer Science and Business Media Deutschland GmbH, 2024. pp. 27-39 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{9c7fe9000ca34d61b1a29a0e9e8ee1b0,

title = "Principal Component Analysis over the Boolean Circuit Within TFHE Scheme",

abstract = "In today{\textquoteright}s information-driven world, the need to protect personal data while maintaining efficient data processing capabilities is crucial. Homomorphic Encryption (HE) has emerged as a potential solution, allowing secure processing of encrypted information without compromising privacy. However, current HE schemes suffer from slow processing speeds, especially when dealing with high-dimensional data. This paper focuses on leveraging the PCA technique within the Fast Fully Homomorphic Encryption over the Torus (TFHE) scheme to optimize the speed of subsequent algorithms. TFHE offers the advantage of enabling the homomorphic implementation of any circuit but suffers from extensive execution time. We present tailored PCA algorithms for TFHE, utilizing the power method and eigen-shift techniques to extract eigenvalues and eigenvectors. These techniques provide efficient solutions for performing PCA computations within the TFHE framework. By designing a dedicated PCA circuit using TFHE{\textquoteright}s fundamental homomorphic gates, we achieve efficient evaluation times for PCA. The performance analysis shows execution times of 3.42 h for a 16-bit dataset and 12.22 h for a 32-bit dataset, with potential for further improvement.",

keywords = "Homomorphic Encryption, Principal Component Analysis, TFHE",

author = "Doh, \{Hyun Jung\} and Yoo, \{Joon Soo\} and Hong, \{Mi Yeon\} and Lee, \{Kang Hoon\} and Ahn, \{Tae Min\} and Yoon, \{Ji Won\}",

note = "Publisher Copyright: {\textcopyright} 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; 24th International Conference on Information Security Applications, WISA 2023 ; Conference date: 23-08-2023 Through 25-08-2023",

year = "2024",

doi = "10.1007/978-981-99-8024-6\_3",

language = "English",

isbn = "9789819980239",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "27--39",

editor = "Howon Kim and Jonghee Youn",

booktitle = "Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers",

address = "Germany",

}

TY - GEN

T1 - Principal Component Analysis over the Boolean Circuit Within TFHE Scheme

AU - Doh, Hyun Jung

AU - Yoo, Joon Soo

AU - Hong, Mi Yeon

AU - Lee, Kang Hoon

AU - Ahn, Tae Min

AU - Yoon, Ji Won

PY - 2024

Y1 - 2024

N2 - In today’s information-driven world, the need to protect personal data while maintaining efficient data processing capabilities is crucial. Homomorphic Encryption (HE) has emerged as a potential solution, allowing secure processing of encrypted information without compromising privacy. However, current HE schemes suffer from slow processing speeds, especially when dealing with high-dimensional data. This paper focuses on leveraging the PCA technique within the Fast Fully Homomorphic Encryption over the Torus (TFHE) scheme to optimize the speed of subsequent algorithms. TFHE offers the advantage of enabling the homomorphic implementation of any circuit but suffers from extensive execution time. We present tailored PCA algorithms for TFHE, utilizing the power method and eigen-shift techniques to extract eigenvalues and eigenvectors. These techniques provide efficient solutions for performing PCA computations within the TFHE framework. By designing a dedicated PCA circuit using TFHE’s fundamental homomorphic gates, we achieve efficient evaluation times for PCA. The performance analysis shows execution times of 3.42 h for a 16-bit dataset and 12.22 h for a 32-bit dataset, with potential for further improvement.

AB - In today’s information-driven world, the need to protect personal data while maintaining efficient data processing capabilities is crucial. Homomorphic Encryption (HE) has emerged as a potential solution, allowing secure processing of encrypted information without compromising privacy. However, current HE schemes suffer from slow processing speeds, especially when dealing with high-dimensional data. This paper focuses on leveraging the PCA technique within the Fast Fully Homomorphic Encryption over the Torus (TFHE) scheme to optimize the speed of subsequent algorithms. TFHE offers the advantage of enabling the homomorphic implementation of any circuit but suffers from extensive execution time. We present tailored PCA algorithms for TFHE, utilizing the power method and eigen-shift techniques to extract eigenvalues and eigenvectors. These techniques provide efficient solutions for performing PCA computations within the TFHE framework. By designing a dedicated PCA circuit using TFHE’s fundamental homomorphic gates, we achieve efficient evaluation times for PCA. The performance analysis shows execution times of 3.42 h for a 16-bit dataset and 12.22 h for a 32-bit dataset, with potential for further improvement.

KW - Homomorphic Encryption

KW - Principal Component Analysis

KW - TFHE

UR - https://www.scopus.com/pages/publications/85182607478

U2 - 10.1007/978-981-99-8024-6_3

DO - 10.1007/978-981-99-8024-6_3

M3 - Conference contribution

AN - SCOPUS:85182607478

SN - 9789819980239

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 27

EP - 39

BT - Information Security Applications - 24th International Conference, WISA 2023, Jeju Island, South Korea, August 23–25, 2023, Revised Selected Papers

A2 - Kim, Howon

A2 - Youn, Jonghee

PB - Springer Science and Business Media Deutschland GmbH

T2 - 24th International Conference on Information Security Applications, WISA 2023

Y2 - 23 August 2023 through 25 August 2023

ER -