VF
            <sup>2</sup>
            Boost: Very Fast Vertical Federated Gradient Boosting for Cross-Enterprise Learning

Fu, Fang; Shao, Yingxia; Yu, Lele; Jiang, Jiawei; Xue, Huanran; Tao, Yangyu; Cui, Bin

doi:10.1145/3448016.3457241

Cited by 43 publications

(17 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this work, we focus on additive HE. For instance, Paillier cryptosystem [53] is a well-known additive HE method and has been used in many FL algorithms [10,19,25,71,75]. Paillier cryptosystem initializes with a key pair ⟨𝑝𝑘, 𝑠𝑘⟩.…”

Section: Privacy-preserving Techniquesmentioning

confidence: 99%

“…Security model. Like many previous works [19,49,66,67,71], we consider the semi-honest (a.k.a. honest-but-curious) security model, where all parties honestly execute the protocols, whilst the curious parties try to analyze the others' data through any information obtained during the protocols.…”

Section: Privacy-preserving Techniquesmentioning

confidence: 99%

“…With the ever-evolving concerns on data privacy, how to build ML models over different data sources with privacy preservation s gaining more popularity. VFL considers the case where different parties jointly train ML models over the partitioned features, which fits numerous real-world cross-enterprise collaborations [10,19,71,74,75]. We categorize the existing works into two lines based on how they process the private features.…”

Section: Vertical Fl (Vfl)mentioning

confidence: 99%

See 2 more Smart Citations

BlindFL: Vertical Federated Machine Learning without Peeking into Your Data

Xue

Cheng

et al. 2022

Proceedings of the 2022 International Conference on Management of Data

Self Cite

View full text Add to dashboard Cite

Due to the rising concerns on privacy protection, how to build machine learning (ML) models over different data sources with security guarantees is gaining more popularity. Vertical federated learning (VFL) describes such a case where ML models are built upon the private data of different participated parties that own disjoint features for the same set of instances, which fits many real-world collaborative tasks. Nevertheless, we find that existing solutions for VFL either support limited kinds of input features or suffer from potential data leakage during the federated execution. To this end, this paper aims to investigate both the functionality and security of ML modes in the VFL scenario.To be specific, we introduce BlindFL, a novel framework for VFL training and inference. First, to address the functionality of VFL models, we propose the federated source layers to unite the data from different parties. Various kinds of features can be supported efficiently by the federated source layers, including dense, sparse, numerical, and categorical features. Second, we carefully analyze the security during the federated execution and formalize the privacy requirements. Based on the analysis, we devise secure and accurate algorithm protocols, and further prove the security guarantees under the ideal-real simulation paradigm. Extensive experiments show that BlindFL supports diverse datasets and models efficiently whilst achieves robust privacy guarantees. CCS CONCEPTS• Computing methodologies → Machine learning; • Security and privacy → Privacy-preserving protocols.

show abstract

Section: Privacy-preserving Techniquesmentioning

confidence: 99%

Section: Privacy-preserving Techniquesmentioning

confidence: 99%

Section: Vertical Fl (Vfl)mentioning

confidence: 99%

See 1 more Smart Citation

BlindFL: Vertical Federated Machine Learning without Peeking into Your Data

Xue

Cheng

et al. 2022

Proceedings of the 2022 International Conference on Management of Data

Self Cite

View full text Add to dashboard Cite

show abstract

“…To address this problem, there are some solutions based on two different technologies: Cryptography, and Local Differential Privacy (LDP). Framework SecureBoost [21] and its subsequent work VF 2 Boost [34] are based on additive homomorphic encryption. Although well-designed engineering optimization is performed, a large number of homomorphic operations still inevitably cause participants to suffer a prohibitively computational overhead.…”

Section: Introductionmentioning

confidence: 99%

“…In SecureBoost proposed by Cheng et al [21], the user holding labels send gradients and hessians encrypted with HE to other users for sorting. Fu et al [34] proposed 𝑉 𝐹 2 Boost to optimize SecureBoost from the perspective of engineering implementation. However, the training process is still extremely time-consuming since a lot of cryptographic operations are irreducible.…”

mentioning

confidence: 99%

OpBoost: A Vertical Federated Tree Boosting Framework Based on Order-Preserving Desensitization

Li¹,

Hu²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Vertical Federated Learning (FL) is a new paradigm that enables users with non-overlapping attributes of the same data samples to jointly train a model without directly sharing the raw data. Nevertheless, recent works show that it's still not sufficient to prevent privacy leakage from the training process or the trained model. This paper focuses on studying the privacy-preserving tree boosting algorithms under the vertical FL. The existing solutions based on cryptography involve heavy computation and communication overhead and are vulnerable to inference attacks. Although the solution based on Local Differential Privacy (LDP) addresses the above problems, it leads to the low accuracy of the trained model.This paper explores to improve the accuracy of the widely deployed tree boosting algorithms satisfying differential privacy under vertical FL. Specifically, we introduce a framework called Op-Boost. Three order-preserving desensitization algorithms satisfying a variant of LDP called distance-based LDP (dLDP) are designed to desensitize the training data. In particular, we optimize the dLDP definition and study efficient sampling distributions to further improve the accuracy and efficiency of the proposed algorithms. The proposed algorithms provide a trade-off between the privacy of pairs with large distance and the utility of desensitized values. Comprehensive evaluations show that OpBoost has a better performance on prediction accuracy of trained models compared with existing LDP approaches on reasonable settings. Our code is open source. 1

show abstract

Practical and General Backdoor Attacks Against Vertical Federated Learning

Xuan,

Chen,

Zhao

et al. 2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

VF ² Boost: Very Fast Vertical Federated Gradient Boosting for Cross-Enterprise Learning

Cited by 43 publications

References 69 publications

BlindFL: Vertical Federated Machine Learning without Peeking into Your Data

BlindFL: Vertical Federated Machine Learning without Peeking into Your Data

OpBoost: A Vertical Federated Tree Boosting Framework Based on Order-Preserving Desensitization

Practical and General Backdoor Attacks Against Vertical Federated Learning

Contact Info

Product

Resources

About

VF 2 Boost: Very Fast Vertical Federated Gradient Boosting for Cross-Enterprise Learning

Cited by 43 publications

References 69 publications

BlindFL: Vertical Federated Machine Learning without Peeking into Your Data

BlindFL: Vertical Federated Machine Learning without Peeking into Your Data

OpBoost: A Vertical Federated Tree Boosting Framework Based on Order-Preserving Desensitization

Practical and General Backdoor Attacks Against Vertical Federated Learning

Contact Info

Product

Resources

About

VF ² Boost: Very Fast Vertical Federated Gradient Boosting for Cross-Enterprise Learning