Secure Computation with Low Communication from Cross-Checking

Gordon, Steven; Ranellucci, Samuel; Wang, Xiao

doi:10.1007/978-3-030-03332-3_3

Cited by 27 publications

(25 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1) Efficient 4PC Protocol: We propose an efficient four-party protocol with active security which proceeds through a masked evaluation inspired by Gordon et al [26]. Our protocol requires 3 ring elements in the online phase per multiplication as opposed to 4 of [26], achieving a 25% improvement. This improvement is achieved by not compromising on the total cost (6 ring elements).…”

Section: A Our Contributionmentioning

confidence: 99%

“…Another significant advantage of our protocol is that the fourth party is not required for evaluation in the online phase. This is not the case with [26], where all the parties need to be online throughout the protocol execution. In addition to the stated contributions, our framework also achieves fairness without affecting the complexity of a multiplication gate.…”

Section: A Our Contributionmentioning

confidence: 99%

“…The training and prediction phases of Linear Regression, Logistic Regression, Neural Networks (NN), and Convolutional Neural Networks (CNN) are used for benchmarking. While we compare our construction with the malicious ABY3 in this section, a more detailed comparison against ABY3 and with the 4PC protocol of Gordon et al [26] appears in Appendix E. a) Environment Details: We provide results for both LAN (1Gbps bandwidth) and WAN (40Mbps bandwidth) settings. In the LAN setting, we have machines with 3.6 GHz Intel Core i7-7700 CPU and 32 GB RAM.…”

Section: Implementation and Benchmarkingmentioning

confidence: 99%

“…The cost for online phase follows directly from Π Mult (Lemma B.4) In this section, we justify the reason for operating in the 4PC setting over the 3PC setting through the lens of benchmarking and practical efficiency. We begin with the comparison of our 4PC secret sharing scheme with that of Gordon et al [26], which shows the reason we built a new 4PC.…”

Section: Cost Comparisonmentioning

confidence: 99%

“…Our ML constructions are built on a new 4PC scheme, instead of the one proposed by Gordon et al [26], primarily due to the following reasons: 1) Our protocol requires only three out of the four parties to be active during most of the online phase. On the contrary, [26] demands all the four parties to be active during the online phase. Thus, our protocol is more efficient in the setting where the computation is outsourced to a set of servers.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Lessons Learned from SunDEW: A Self Defense Environment for Web Applications

Şahin¹,

Hebert²,

Oliveira³

2020

Proceedings 2020 Workshop on Measurements, Attacks, and Defenses for the Web

View full text Add to dashboard Cite

Machine learning has started to be deployed in fields such as healthcare and finance, which involves dealing with a lot of sensitive data. This propelled the need for and growth of privacy-preserving machine learning (PPML). We propose an actively secure four-party protocol (4PC), and a framework for PPML, showcasing its applications on four of the most widely-known machine learning algorithms -Linear Regression, Logistic Regression, Neural Networks, and Convolutional Neural Networks. (Communication). Protocol Π BitExt (Fig. 19) requires 1 round and a communication of 4 + 1 bits in the offline phase, while it requires 3 rounds and a communication of 5 + 2 bits in the online phase.Proof: During the offline phase, parties execute one instance each of Π vSh and Π B vSh resulting in one round and a communication of + 1 bits (Lemma C.1). Also, the offline phase for multiplication is performed resulting in an additional communication of 3 bits.During the online phase, parties first execute an arithmetic multiplication, resulting in one round and communication of 3 bits (Lemma B.4). The value rv is reconstructed towards both P 0 and P 3 , resulting in an additional round and an amortized communication of 2 bits. This is followed by the last round, where parties execute one instance of Π B vSh resulting in a communication of 2 bits. Thus the online phase requires three rounds and an amortized communication of 5 + 2 bits.Lemma D.4 (Communication). Protocol ReLU (Π relu ) requires 3 rounds and a communication of 8 + 2 bits in the offline phase, while it requires 4 rounds and a communication of 8 +2 bits in the online phase.

show abstract

Section: A Our Contributionmentioning

confidence: 99%

Section: A Our Contributionmentioning

confidence: 99%

Section: Implementation and Benchmarkingmentioning

confidence: 99%

Section: Cost Comparisonmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations