Efficient Bit-Decomposition and Modulus-Conversion Protocols with an Honest Majority

Kikuchi, Ryo; Ikarashi, Dai; Matsuda, Takahiro; Hamada, Keisuke; Chida, Koji

doi:10.1007/978-3-319-93638-3_5

Cited by 9 publications

(29 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ability to compute q (in shared form), which we refer to as quotient transfer, will play an important role in our division protocol. Kikuchi et al [28] proposed efficient three-party quotient transfer protocols for passive and active security. In this paper, we use their protocols specialized to the following setting; firstly, we use a Mersenne prime p for the field Z p underlying the sharings, and we will use • -sharing for passive security and • -sharing for active security.…”

Section: Quotient Transfer Protocolmentioning

confidence: 99%

“…It means that a secret a should satisfy 2a < p and 4a < p for passive and active security, respectively. These protocols have been implicitly used as building blocks for other protocols in [28], but were not explicitly defined. We thus describe them in Appendix B, as well as a quotient transfer protocol for [•]-sharings, which is a popular way to obtain the carry for binary addition.…”

Section: Quotient Transfer Protocolmentioning

confidence: 99%

“…The insight behind our protocols is that the qα p term can be eliminated, which we essentially achieve via the quotient transfer protocol [28], that allow us to obtain q efficiently. This protocol suits our setting since it requires a prime p, and prefers a Mersenne prime.…”

Section: B Our Protocol For Division By Public Value 1) Intuitionmentioning

confidence: 99%

“…The quotient transfer protocol in [28] requires 2 bits of communication and 1 communication round, besides a single call of F mod . The modulus conversion protocol in [28] and ConvertToRep require 3|p| + 3 bits and 2|p| bits of communication, respectively, and both 1 communication round. Furthermore, we can reduce the number of rounds required in Protocol 2 by parallel execution 2 of F QT and ConvertToRep.…”

Section: ) Efficiencymentioning

confidence: 99%

“…Furthermore, we can reduce the number of rounds required in Protocol 2 by parallel execution 2 of F QT and ConvertToRep. Consequently, instantiating F QT (and F mod used in QT internally) by the protocols in [28], Protocol 1 and 2 require 3|p|+5 and 5|p| + 5 bits of communication and 2 communications rounds in total.…”

Section: ) Efficiencymentioning

confidence: 99%

See 4 more Smart Citations

Adam in Private: Secure and Fast Training of Deep Neural Networks with Adaptive Moment Estimation

Attrapadung¹,

Hamada²,

Ikarashi³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Machine Learning (ML) algorithms, especially deep neural networks (DNN), have proven themselves to be extremely useful tools for data analysis, and are increasingly being deployed in systems operating on sensitive data, such as recommendation systems, banking fraud detection, and healthcare systems. This underscores the need for privacy-preserving ML (PPML) systems, and has inspired a line of research into how such systems can be constructed efficiently. We contribute to this line of research by proposing a framework that allows efficient and secure evaluation of full-fledged state-of-the-art ML algorithms via secure multiparty computation (MPC). This is in contrast to most prior works on PPML, which require advanced ML algorithms to be substituted with approximated variants that are "MPC-friendly", before MPC techniques are applied to obtain a PPML algorithm. A drawback of the latter approach is that it requires careful finetuning of the combined ML and MPC algorithms, and might lead to less efficient algorithms or inferior quality ML (such as lower prediction accuracy). This is an issue for secure training of DNNs in particular, as this involves several arithmetic algorithms that are thought to be "MPC-unfriendly", namely, integer division, exponentiation, inversion, and square root extraction.In this work, we propose secure and efficient protocols for the above seemingly MPC-unfriendly computations (but which are essential to DNN). Our protocols are three-party protocols in the honest-majority setting, and we propose both passively secure and actively secure with abort variants. A notable feature of our protocols is that they simultaneously provide high accuracy and efficiency. This framework enables us to efficiently and securely compute modern ML algorithms such as Adam (Adaptive moment estimation) and the softmax function "as is", without resorting to approximations. As a result, we obtain secure DNN training that outperforms state-of-the-art three-party systems; our full training is up to 6.7 times faster than just the online phase of the recently proposed FALCON (Wagh et al. at PETS'21) on the standard benchmark network for secure training of DNNs. To further demonstrate the scalability of our protocols, we perform measurements on real-world DNNs, AlexNet and VGG16, which are complex networks containing millions of parameters. The performance of our framework for these networks is up to a factor of about 12 ∼ 14 faster for AlexNet and 46 ∼ 48 faster for VGG16 to achieve an accuracy of 70% and 75%, respectively, when compared to FALCON.

show abstract

Section: Quotient Transfer Protocolmentioning

confidence: 99%

Section: Quotient Transfer Protocolmentioning

confidence: 99%

Section: B Our Protocol For Division By Public Value 1) Intuitionmentioning

confidence: 99%

Section: ) Efficiencymentioning

confidence: 99%

Section: ) Efficiencymentioning

confidence: 99%

See 3 more Smart Citations