Fairness, integrity, and privacy in a scalable blockchain-based federated learning system

Rückel, Timon; Sedlmeir, Johannes; Hofmann, Peter

doi:10.1016/j.comnet.2021.108621

Cited by 36 publications

(38 citation statements)

References 72 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our protocol uses a decentralized pseudorandom function (such as a verifiable on-chain oracles 3 , or those used in [5]) to provide a random seed used to select computers based on reputation. This removes the possibility of reputation-based manipulation by clusters of colluding computers who may try to prioritize the selection of computers in the colluding cluster, as is allowed in existing works [8,7,24].…”

Section: Our Contributionmentioning

confidence: 99%

“…Although our privacy techniques are not novel outside of DC, and in fact reduce privacy compared to DC protocols like [24], the combination of decentralisation, proven strong incentive compatibility and the ability to apply one smart-contract instance of Privacy Marvel DC to any computational problem with output in Euclidean space (summarized in Table 1) stands as an additional novel contribution.…”

Section: Our Contributionmentioning

confidence: 99%

“…Strong Incentive Compatible Computation-Independent Diminishing Adversarial Selection Prob. [16,1,20,19,13,27,26] / Toyoda et al [25] Ruckel et al [24] *…”

Section: Protocol Tokenized Rewardsmentioning

confidence: 99%

“…In decentralized settings such as blockchain protocols, players only follow an action(s) if that action(s) is(are) strong incentive compatible, resulting in strictly higher payoffs than the alternatives. One method to combat this is, for a given computation, to use zero-knowledge (ZK) tools to prove that a computer performed the actions as prescribed by the requester [24]. Although theoretically any computation can be encoded in this way, the practicality of requiring requesters, typically with low computing power by the nature of outsourcing, to encode their computation as a ZK-circuit to allow for the proving of correct computation is an open question.…”

Section: Introductionmentioning

confidence: 99%

“…Although theoretically any computation can be encoded in this way, the practicality of requiring requesters, typically with low computing power by the nature of outsourcing, to encode their computation as a ZK-circuit to allow for the proving of correct computation is an open question. Furthermore, the pre-computation work required by requesters in [24] is itself intensive (3-40 minutes using 16 virtual CPU cores, and 64 GB RAM), and not appropriate for computationally-restricted requesters, defeating a significant purpose of DC protocols. In this paper, we address all of these issues.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Marvel DC: A Blockchain-Based Decentralized and Incentive-Compatible Distributed Computing Protocol

McMenamin¹,

Daza²

2022

Preprint

View full text Add to dashboard Cite

Decentralized computation outsourcing should allow anyone to access the large amounts of computational power that exists in the Internet of Things.Unfortunately, when trusted third parties are removed to achieve this decentralization, ensuring an outsourced computation is performed correctly remains a significant challenge. In this paper, we provide a solution to this problem.We outline Marvel DC, a fully decentralized blockchain-based distributedcomputing protocol which formally guarantees that computers are strictly incentivized to correctly perform requested computations. Furthermore, Marvel DC utilizes a reputation management protocol to ensure that, for any minority of computers not performing calculations correctly, these computers are identified and selected for computations with diminishing probability. We then outline Privacy Marvel DC, a privacy-enhanced version of Marvel DC which decouples results from the computers which computed them, making the protocol suitable for computations such as Federated Learning, where results can reveal sensitive information about that computer that computed them. We provide an implementation of Marvel DC and analyses of both protocols, demonstrating that This document is part of a project that has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement number 814284

show abstract

Section: Our Contributionmentioning

confidence: 99%

Section: Our Contributionmentioning

confidence: 99%

“…Strong Incentive Compatible Computation-Independent Diminishing Adversarial Selection Prob. [16,1,20,19,13,27,26] / Toyoda et al [25] Ruckel et al [24] *…”

Section: Protocol Tokenized Rewardsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Marvel DC: A Blockchain-Based Decentralized and Incentive-Compatible Distributed Computing Protocol

McMenamin¹,

Daza²

2022

Preprint

View full text Add to dashboard Cite

show abstract

Privacy preserving and secure robust federated learning: A survey

Han,

Lu,

Wang

et al. 2024

Concurrency and Computation

View full text Add to dashboard Cite

SummaryFederated learning (FL) has emerged as a promising solution to address the challenges posed by data silos and the need for global data fusion. It offers a distributed machine learning framework with privacy‐preserving features, allowing model training without the need to collect user data. However, FL also presents significant security and privacy threats that hinder its widespread adoption. The requirements of privacy and security in FL are inherently conflicting. Privacy necessitates the concealment of individual client updates, while security requires the disclosure of client updates to detect anomalies. While most existing research focused on the privacy and security aspects of FL, very few studies have addressed the compatibility of these two demands. In this work, we aim to bridge this gap by proposing a comprehensive defense scheme that ensures privacy, security, and compatibility in FL. We categorize the existing literature into two key directions: privacy defense and security defense. Privacy defense includes methods based on additive masks, differential privacy, homomorphic encryption, and trusted execution environment, whereas security defense encompasses distance‐, performance‐, clustering‐, and similarity‐based anomaly detection techniques and statistical information‐based anomaly update bypassing techniques when the server is trusted and privacy‐compatible anomaly update detection techniques when the server is not trusted. In addition, this article presents decentralized FL solutions based on blockchain. For each direction, we discuss specific technical solutions, their advantages, and disadvantages. By evaluating various defense methods, we identify the most suitable approach to address the primary challenge of “achieving a secure and robust FL system against malicious adversaries while protecting users' privacy.” We then propose a theoretical reference framework for end‐to‐end protection of privacy and security in FL for the key problem, which summarizes the attack surface of FL systems from the client to the server under the security model where the client and server are malicious. Leveraging the strengths and characteristics of existing schemes, our proposed framework integrates multiple techniques to strike a balance between privacy, usability, and efficiency. This framework serves as a valuable reference and provides insights for future work in the field. Finally, we also provide recommendations for future research directions in this field.

show abstract

Systematic review: Comparing zk‐SNARK, zk‐STARK, and bulletproof protocols for privacy‐preserving authentication

Oude Roelink,

El‐Hajj,

Sarmah

2024

Security and Privacy

View full text Add to dashboard Cite

This systematic literature review examines the implementation and analysis of zk‐SNARK, zk‐STARK, and bulletproof non‐interactive zero‐knowledge proof (NIZKP) protocols in privacy‐preserving applications across diverse sectors. Examining 41 research works obtained through the systematic search queries and filtering criteria published from 2015 to April 2023, we categorized findings into financial, medical, business, general, and other domains. Our analysis highlights significant variations of up to several orders of magnitude in real‐world performance across implementations utilizing NIZKP protocols. However, divergent methodologies in security analyses hindered conclusive comparisons. Addressing research gaps, our future endeavors aim to establish a real‐world benchmark for these protocols.

show abstract

Fairness, integrity, and privacy in a scalable blockchain-based federated learning system

Cited by 36 publications

References 72 publications

Marvel DC: A Blockchain-Based Decentralized and Incentive-Compatible Distributed Computing Protocol

Marvel DC: A Blockchain-Based Decentralized and Incentive-Compatible Distributed Computing Protocol

Privacy preserving and secure robust federated learning: A survey

Systematic review: Comparing zk‐SNARK, zk‐STARK, and bulletproof protocols for privacy‐preserving authentication

Contact Info

Product

Resources

About