Federated Learning With Differential Privacy: Algorithms and Performance Analysis

Wei, Kang; Li, Jun; Ding, Ming; Ma, Chuan; Yang, Howard H.; Farokhi, Farhad; Jin, Shi; Quek, Tony Q. S.; Poor, H. Vincent

doi:10.1109/tifs.2020.2988575

Cited by 1,219 publications

(536 citation statements)

References 23 publications

Supporting

Mentioning

434

Contrasting

Order By: Relevance

“…Federated learning can be integrated with differential privacy techniques [234] to establish a stronger standard for individual privacy in modelling tasks on sensitive FRM datasets [235]. Differential privacy has been widely accepted by both researchers and practitioners as the de facto standard of privacy [236].…”

Section: Federated Learningmentioning

confidence: 99%

Machine Learning for Financial Risk Management: A Survey

et al. 2020

View full text Add to dashboard Cite

Financial risk management avoids losses and maximizes profits, and hence is vital to most businesses. As the task relies heavily on information-driven decision making, machine learning is a promising source for new methods and technologies. In recent years, we have seen increasing adoption of machine learning methods for various risk management tasks. Machine-learning researchers, however, often struggle to navigate the vast and complex domain knowledge and the fast-evolving literature. This paper fills this gap, by providing a systematic survey of the rapidly growing literature of machine learning research for financial risk management. The contributions of the paper are four-folds: First, we present a taxonomy of financial-risk-management tasks and connect them with relevant machine learning methods. Secondly, we highlight significant publications in the past decade. Thirdly, we identify major challenges being faced by researchers in this area. And finally, we point out emerging trends and promising research directions.

show abstract

Section: Federated Learningmentioning

confidence: 99%

Machine Learning for Financial Risk Management: A Survey

et al. 2020

View full text Add to dashboard Cite

show abstract

“…• FL is a subset of MPC from the m-ary functionality point of view; • If FL attains the security in the simulation-based framework, then the resulting SFL is a subset of SMPC. Since there are other known techniques such as the homomorphic encryption (HE) and the dierential privacy (DP) to attain the privacy of federated learning procedures [6], [9] and we are not clear whether SFL is a subset of HE or DP within the correspondent framework, we thus leave these interesting problems to the research community.…”

Section: B Our Contributionmentioning

confidence: 99%

“…The datasets dened in the FL framework can be categorized as horizontal, vertical and hybrid types. Roughly speaking, in the horizontal FL, the feature spaces of datasets among dierent organizations (data owners) are same but not overlapped over the sample spaces [5]; in the vertical FL, the sample spaces of datasets among dierent organizations are same but not overlapped over the feature spaces [6], [7]; in the hybrid FL, both feature spaces and sample spaces of dierent organizations are overlapped [8], [9]. We refer to the reader [10] [14] and the references therein for more details.…”

Section: Introductionmentioning

confidence: 99%

Privacy-Preserving Weighted Federated Learning Within the Secret Sharing Framework

Zhu

Goh

2020

IEEE Access

View full text Add to dashboard Cite

This paper studies privacy-preserving weighted federated learning within the secret sharing framework, where individual private data is split into random shares which are distributed among a set of pre-dened computing servers. The contribution of this paper mainly comprises the following four-fold: • In the rst fold, the relationship between federated learning (FL) and multi-party computation (MPC) as well as that of secure federated learning (SFL) and secure multi-party computation (SMPC) is investigated. We show that FL is a subset of MPC from the m-ary functionality point of view. Furthermore, if the underlying FL instance privately computes the dened mary functionality in the simulation-based framework, then the simulation-based FL solution is an instance of SMPC. • In the second fold, a new notion which we call weighted federated learning (wFL) is introduced and formalized. Then an oracle-aided SMPC for computing wFL is presented and analysed by decoupling the security of FL from that of MPC. Our decoupling formulation of wFL benets FL developers selecting their best security practices from the state-of-the-art security tools. • In the third-fold, a concrete implementation of wFL leveraging the random splitting technique in the framework of the 3-party computation is presented and analysed. The security of our implementation is guaranteed by the security composition theorem within the secret share framework. • In the fourth-fold, a complement to MASCOT is introduced and formalized in the framework of SPDZ, where a novel solution to the Beaver triple generator is constructed from the standard El Gamal encryption. Our solution is formalized as a three-party computation and a generation of the Beaver triple requires roughly 5 invocations of the El Gamal encryptions. We are able to show that the proposed implementation is secure against honest-but-curious adversary assuming that the underlying El Gamal encryption is semantically secure.

show abstract

“…Especially, a variety of IoT applications call for data mining and learning securely and reliably in distributed systems; it represents a vision in which the Internet extends into the real world while embracing everyday objects. This new trend arises from synergically merging IoT and distributed computing [1]- [2].…”

Section: Introductionmentioning

confidence: 99%

“…In order to solve this issue, it is necessary to add artificial noise to preventing personal information leakage; one prominent example of which is the concept of differential privacy (DP). The major goal of DP in the FL process is to ensure that a learned model does not reveal whether a certain data point was used during training [2], [4], [7].…”

Section: Introductionmentioning

confidence: 99%

Incentive Design and Differential Privacy Based Federated Learning: A Mechanism Design Perspective

Kim

2020

IEEE Access

View full text Add to dashboard Cite

Due to stricter data management regulations and large size of the training data, distributed learning paradigm such as federated learning (FL) has gained attention recently. FL is capable of significantly preserving end-users' private data from being exposed to external adversaries. However, private information can still be divulged by uploading parameters from users. Therefore, a key challenge in the FL platform is how users participate to build a high-quality learning model with effectively preventing information leakage. To address the above challenge, we design a novel incentive mechanism to attract more data owners to join in the FL process with the consideration of privacy preservation. To implement our proposed scheme, we adopt the concepts of mechanism design (MD) and differential privacy (DP); MD takes an objectives-first approach to designing incentives toward desired objectives, and the DP can provide a theoretical guarantee for users' privacy in FL participations. Based on the DP based incentive mechanism, our joint approach can leverage the full synergy that gives mutual advantages for users and learning operators. Therefore, we can take various benefits in a rational way under the dynamic changing FL environments. Through simulation analysis, the numerical results validate the performance efficiency of our proposed scheme.

show abstract

Federated Learning With Differential Privacy: Algorithms and Performance Analysis

Cited by 1,219 publications

References 23 publications

Machine Learning for Financial Risk Management: A Survey

Machine Learning for Financial Risk Management: A Survey

Privacy-Preserving Weighted Federated Learning Within the Secret Sharing Framework

Incentive Design and Differential Privacy Based Federated Learning: A Mechanism Design Perspective

Contact Info

Product

Resources

About