Presentation 2022-07-28
Acceleration of HE-Transformer with bit reduced SEAL and HEXL
Xinyi Li, Masaki Nishi, Teppei Shishido, Keiji Kimura,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) While cloud applications are covering people's privacy data, their related risks, such as theft risk, leak risk, and other serious risks, are becoming critical issues. Homomorphic encryption (HE) is a promising approach to overcome them since it enables calculation of encrypted data without decryption. Deep learning (DL) inference on cloud is an interesting application of HE. However, the computational cost of HE is too expensive, resulting in its limited usage in the real world. In this paper, we propose an acceleration technique of DL on HE, and implement it in HE-Transformer, which is a framework for DL on HE. Our approach is reducing the bit-width of internal data structure of HE libraries, SEAL and HEXL, from 64-bit to 32-bit. As a result, we can obtain up to $3.59times$ speedup compared to the original version, while still keeping data privacy.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Homomorphic Encryption / Deep Learning / SIMD
Paper # CPSY2022-12,DC2022-12
Date of Issue 2022-07-20 (CPSY, DC)

Conference Information
Committee CPSY / DC / IPSJ-ARC
Conference Date 2022/7/27(3days)
Place (in Japanese) (See Japanese page)
Place (in English) Kaikyo Messe Shimonoseki
Topics (in Japanese) (See Japanese page)
Topics (in English) SWoPP2022: Parallel, Distributed and Cooperative Processing Systems and Dependable Computing
Chair Michihiro Koibuchi(NII) / Tatsuhiro Tsuchiya(Osaka Univ.) / 津邑 公暁(名工大)
Vice Chair Kota Nakajima(Fujitsu Lab.) / Tomoaki Tsumura(Nagoya Inst. of Tech.) / Toshinori Hosokawa(Nihon Univ.)
Secretary Kota Nakajima(JAIST) / Tomoaki Tsumura(Hitachi) / Toshinori Hosokawa(Nihon Univ.) / (Chiba Univ.)
Assistant Ryohei Kobayashi(Tsukuba Univ.) / Takaaki Miyajima(Meiji Univ.)

Paper Information
Registration To Technical Committee on Computer Systems / Technical Committee on Dependable Computing / Special Interest Group on System Architecture
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Acceleration of HE-Transformer with bit reduced SEAL and HEXL
Sub Title (in English)
Keyword(1) Homomorphic Encryption
Keyword(2) Deep Learning
Keyword(3) SIMD
1st Author's Name Xinyi Li
1st Author's Affiliation Waseda University(Waseda Univ.)
2nd Author's Name Masaki Nishi
2nd Author's Affiliation Waseda University(Waseda Univ.)
3rd Author's Name Teppei Shishido
3rd Author's Affiliation Waseda University(Waseda Univ.)
4th Author's Name Keiji Kimura
4th Author's Affiliation Waseda University(Waseda Univ.)
Date 2022-07-28
Paper # CPSY2022-12,DC2022-12
Volume (vol) vol.122
Number (no) CPSY-133,DC-134
Page pp.pp.65-70(CPSY), pp.65-70(DC),
#Pages 6
Date of Issue 2022-07-20 (CPSY, DC)