Secure Publish-Process-Subscribe System for Dispersed Computing

Jin, Weizhao; Krishnamachari, Bhaskar; Naveed, Muhammad; Ravi, Srivatsan; Sanou, Eduard; Wright, Kwame-Lante

doi:10.1109/srds55811.2022.00016

Cited by 1 publication

(2 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Currently, several open-source HE libraries are available [23, 40,44]. Among these libraries, PALISADE has one of the fastest HE implementations [20] (verified by our experimental results in §5.2.5) and also supports several decentralized multi-party functionalities, such as Proxy Re-Encryption [5,28] and Threshold Homomorphic Encryption [4] for serverless or decentralized FL scenarios [22]. In our framework, we prioritize PALISADE as our HE core.…”

Section: He Implementationmentioning

confidence: 99%

“…HE-based federated aggregation encrypts local models during aggregation and performs model aggregation over ciphertexts. This approach can enable secure federated learning deployments and has been adopted by several FL systems [15,26,38,57] and a few domain-specific applications have been implemented [28,47,56 Despite advances in the Homomorphic Encryption research community to enhance HE performance in theory and practice, HE remains a powerful but complex cryptographic foundation with impractical overheads for most real-world applications. Prior HE solutions mainly employ existing generic HE methods without sufficient optimization for large-scale privacy-preserving federated learning deployment.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System

Jin¹,

Yao²,

Han³

et al. 2023

Preprint

View full text Add to dashboard Cite

Federated Learning (FL) enables machine learning model training on distributed edge devices by aggregating local model updates rather than local data. However, privacy concerns arise as the FL server's access to local model updates can potentially reveal sensitive personal information by performing attacks like gradient inversion recovery. To address these concerns, privacypreserving methods, such as Homomorphic Encryption (HE)-based approaches, have been proposed. Despite HE's post-quantum security advantages, its applications suffer from impractical overheads. In this paper, we present FedML-HE, the first practical system for efficient HE-based secure federated aggregation that provides a user/device-friendly deployment platform. FedML-HE utilizes a novel universal overhead optimization scheme, significantly reducing both computation and communication overheads during deployment while providing customizable privacy guarantees. Our optimized system demonstrates considerable overhead reduction, particularly for large models (e.g., ∼10x reduction for HE-federated training of ResNet-50 and ∼40x reduction for BERT), demonstrating the potential for scalable HEbased FL deployment.

show abstract