Non-interactive verifiable privacy-preserving federated learning

Xu, Yi; Peng, Changgen; Tan, Weijie; Tian, Youliang; Ma, Minyao; Niu, Kun

doi:10.1016/j.future.2021.10.017

Cited by 22 publications

(8 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Bonawitz et al [38] built a privacy-preserving FL framework, where SMC can securely aggregate the clients' parameters, and be robust to clients' exit. Xu et al [39] proposed a non-interactive and verifiable privacy protection FL model, and proposed a novel privacy gradient aggregation scheme using random matrix coding and secure two-party computation. These SMC models lead to huge communication cost because of multiple interactions involving in the learning process.…”

Section: Secure Multi-party Computingmentioning

confidence: 99%

“…Cause huge communication cost in multi-party interaction [36] Use SMC technology to achieve a robust reversible image watermarking scheme [37] Protect the private information of specific location in privacy-aware indoor location application [38] Can securely aggregate the clients' parameters in FL [39] Propose a novel privacy gradient aggregation scheme using random matrix coding and secure two-party computation Differential privacy [40] Achieve a balance between privacy loss and model performance while hiding customer contributions and private data during training.…”

Section: Reference Contributionmentioning

confidence: 99%

See 1 more Smart Citation

A survey on federated learning: challenges and applications

Zhang

Lan

et al. 2022

Int. J. Mach. Learn. & Cyber.

102

View full text Add to dashboard Cite

y impact on human life [1][2][3]. As great progress has been made in AI technology in recent years, various application fields have stepped into intelligence [4,5]. On the road of AI development, models, computing power, chip performance, and other technical issues have been the focus of academic research, so that AI technology can continue to evolve. For machines to truly approach the level of human thought, they need to be trained with vast amounts of real data [6,7]. However, cloud computing power, data security, data silos and other risks will inevitably become constraints for AI to win user trust, collect private data, and achieve largescale implementation [8]. Therefore, it is an urgent need for a practical and effective technique to alleviate the above problems and make the AI full of vitality again. Under this background, the concept of "federated learning (FL) " came into being.The notion of FL is first proposed by Google in 2016, mainly to make android mobile phone users update models locally without revealing private personal data [9]. After then, Google implemented an application-oriented FL system. The designed FL system, which focused on running federated average (FedAvg) algorithms on mobile phones, can perform federated analytics and be applied to monitor statistics for large-scale cluster equipment without recording raw device data to the cloud server. FL is one of the most Zhihua Cui

show abstract

Section: Secure Multi-party Computingmentioning

confidence: 99%

Section: Reference Contributionmentioning

confidence: 99%

A survey on federated learning: challenges and applications

Zhang

Lan

et al. 2022

Int. J. Mach. Learn. & Cyber.

102

View full text Add to dashboard Cite

show abstract

“…In Fastsecagg [147], by sacrificing some security, the authors substitute the standard Shamir secret sharing with a more efficient FFT-based multisecret sharing scheme. Alternatively, models can be shared between two servers as in [148], [149], [150], [151], or several servers such as [152], [153]. Some other works introduce a two-phase secret-sharing-based aggregation [154], [155].…”

Section: Mpc-based Aggregationmentioning

confidence: 99%

Privacy-Preserving Aggregation in Federated Learning: A Survey

Liu¹,

Guo²,

Yang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Over the recent years, with the increasing adoption of Federated Learning (FL) algorithms and growing concerns over personal data privacy, Privacy-Preserving Federated Learning (PPFL) has attracted tremendous attention from both academia and industry. Practical PPFL typically allows multiple participants to individually train their machine learning models, which are then aggregated to construct a global model in a privacy-preserving manner. As such, Privacy-Preserving Aggregation (PPAgg) as the key protocol in PPFL has received substantial research interest. This survey aims to fill the gap between a large number of studies on PPFL, where PPAgg is adopted to provide a privacy guarantee, and the lack of a comprehensive survey on the PPAgg protocols applied in FL systems. In this survey, we review the PPAgg protocols proposed to address privacy and security issues in FL systems. The focus is placed on the construction of PPAgg protocols with an extensive analysis of the advantages and disadvantages of these selected PPAgg protocols and solutions. Additionally, we discuss the open-source FL frameworks that support PPAgg. Finally, we highlight important challenges and future research directions for applying PPAgg to FL systems and the combination of PPAgg with other technologies for further security improvement.

show abstract

“…The sums of honest clients' inputs are same in both worlds. The encrypted data [x ̂𝑖] 𝑝𝑘 is indistinguishable due to the semantic security of DP encryption scheme [7]. The signature on indistinguishable encrypted data is also indistinguishable.…”

Section: Full Version Of Nipvs-fl Initialization Phase (Kgc)mentioning

confidence: 99%

“…This is a strong assumption that needs to be improved. Xu et al [7] avoid using the popular basic aggregation scheme which utilizing the pairwise additive masking method. They design a dual-servers architecture, and consider using random matrix coding to encrypt the gradient matrix.…”

Section: Introductionmentioning

confidence: 99%

NIPVS-FL: a non-interactive publicly verifiable secure federated-learning scheme against malicious servers

Wang

Cao²,

Dong³

et al. 2022

Third International Conference on Computer Communication and Network Security (CCNS 2022)

View full text Add to dashboard Cite

Federated learning (FL) enables decentralized data sources like mobile phones to joint training a neural network model without sharing the original data. However, shared local gradients make the privacy of local data in FL vulnerable. The aggregation server also may return incorrect results to clients due to unexpected error or the deliberately attack. In this work, we explore how to design a non-interactive and publicly verifiable aggregation scheme. The existing verifiable schemes are under semi-honest adversary model, in which the server is honest-but-curious but with additional power to counterfeit the aggregation result. We propose a scheme under stronger security model against malicious servers. The proposed scheme guarantees that as long as the two servers are non-colluding, even a malicious server cannot obtain input privacy of client. The malicious server will be detected by honest clients when it tries to tamper the result.

show abstract

Non-interactive verifiable privacy-preserving federated learning

Cited by 22 publications

References 23 publications

A survey on federated learning: challenges and applications

A survey on federated learning: challenges and applications

Privacy-Preserving Aggregation in Federated Learning: A Survey

NIPVS-FL: a non-interactive publicly verifiable secure federated-learning scheme against malicious servers

Contact Info

Product

Resources

About