Secure Floating-Point Training

被引：0

作者：

Rathee, Deevashwer ^{[1
]}

Bhattacharya, Anwesh ^{[2
]}

Gupta, Divya ^{[2
]}

Sharma, Rahul ^{[2
]}

Song, Dawn ^{[1
]}

机构：

[1] Univ Calif Berkeley, Berkeley, CA 94704 USA

[2] Microsoft Res, Redmond, WA USA

来源：

PROCEEDINGS OF THE 32ND USENIX SECURITY SYMPOSIUM | 2023年

关键词：

D O I：

暂无

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Secure 2-party computation (2PC) of floating-point arithmetic is improving in performance and recent work runs deep learning algorithms with it, while being as numerically precise as commonly used machine learning (ML) frameworks like PyTorch. We find that the existing 2PC libraries for floating-point support generic computations and lack specialized support for ML training. Hence, their latency and communication costs for compound operations (e.g., dot products) are high. We provide novel specialized 2PC protocols for compound operations and prove their precision using numerical analysis. Our implementation BEACON outperforms state-of-the-art libraries for 2PC of floating-point by over 6x.

引用

页码：6329 / 6346

页数：18

共 50 条

[1] Secure multiparty computations in floating-point arithmetic
Guo, Chuan
Hannun, Awni
Knott, Brian
van der Maaten, Laurens
Tygert, Mark
Zhu, Ruiyu
INFORMATION AND INFERENCE-A JOURNAL OF THE IMA, 2022, 11 (01) : 103 - 135
[2] Performance Analysis of Secure Floating-Point Sums and Dot Products
Catrina, Octavian
2020 13TH INTERNATIONAL CONFERENCE ON COMMUNICATIONS (COMM), 2020, : 465 - 470
[3] Complexity and Performance of Secure Floating-Point Polynomial Evaluation Protocols
Catrina, Octavian
COMPUTER SECURITY - ESORICS 2021, PT II, 2021, 12973 : 352 - 369
[4] Secure computation of hidden Markov models and secure floating-point arithmetic in the malicious model
Mehrdad Aliasgari
Marina Blanton
Fattaneh Bayatbabolghani
International Journal of Information Security, 2017, 16 : 577 - 601
[5] Accurate Floating-point Operation using Controlled Floating-point Precision
Zaki, Ahmad M.
Bahaa-Eldin, Ayman M.
El-Shafey, Mohamed H.
Aly, Gamal M.
2011 IEEE PACIFIC RIM CONFERENCE ON COMMUNICATIONS, COMPUTERS AND SIGNAL PROCESSING (PACRIM), 2011, : 696 - 701
[6] Secure computation of hidden Markov models and secure floating-point arithmetic in the malicious model
Aliasgari, Mehrdad
Blanton, Marina
Bayatbabolghani, Fattaneh
INTERNATIONAL JOURNAL OF INFORMATION SECURITY, 2017, 16 (06) : 577 - 601
[7] OPTIMIZING SECURE FLOATING-POINT ARITHMETIC: SUMS, DOT PRODUCTS, AND POLYNOMIALS
Catrina, Octavian
PROCEEDINGS OF THE ROMANIAN ACADEMY SERIES A-MATHEMATICS PHYSICS TECHNICAL SCIENCES INFORMATION SCIENCE, 2020, 21 (01): : 21 - 28
[8] Floating-point arithmetic
Boldo, Sylvie
Jeannerod, Claude-Pierre
Melquiond, Guillaume
Muller, Jean-Michel
ACTA NUMERICA, 2023, 32 : 203 - 290
[9] Optimization and Tradeoffs in Secure Floating-Point Computation: Products, Powers, and Polynomials
Catrina, Octavian
PROCEEDINGS OF THE 6TH CONFERENCE ON THE ENGINEERING OF COMPUTER BASED SYSTEMS (ECBS 2019), 2020,
[10] Efficient Secure Floating-point Arithmetic using Shamir Secret Sharing
Catrina, Octavian
PROCEEDINGS OF THE 16TH INTERNATIONAL JOINT CONFERENCE ON E-BUSINESS AND TELECOMMUNICATIONS, VOL 2: SECRYPT, 2019, : 49 - 60

← 1 2 3 4 5 →