Byzantine-Tolerant Machine Learning

Blanchard, Peva; Mhamdi, El Mahdi El; Guerraoui, Rachid; Stainer, Julien

doi:10.48550/arxiv.1703.02757

Cited by 15 publications

(39 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We note that with our construction, all the aggregation results {A r } r∈P from the successful rounds are independently and identically distributed (since each data owner performs local computations with M 0 data points, and each round aggregates results from N DOs). Therefore, according to Proposition 2 and Proposition 3 in (Blanchard et al, 2017b), as long as |P | < (1 − 2µ)|P| − 2, where µ is the maximum fraction of the aggregation results that may be corrupted, the estimated overall gradient in (10) provides a close approximation of the true gradient, which leads to the convergence of the model training.…”

Section: Security Of Model Updatementioning

confidence: 97%

See 1 more Smart Citation

OmniLytics: A Blockchain-based Secure Data Market for Decentralized Machine Learning

Liang

Jiang

2021

Preprint

View full text Add to dashboard Cite

We propose OmniLytics, a blockchain-based secure data trading marketplace for machine learning applications. Utilizing OmniLytics, many distributed data owners can contribute their private data to collectively train a ML model requested by some model owners, and get compensated for data contribution. OmniLytics enables such model training while simultaneously providing 1) model security against curious data owners; 2) data security against curious model and data owners; 3) resilience to malicious data owners who provide faulty results to poison model training; and 4) resilience to malicious model owner who intents to evade the payment. OmniLytics is implemented as a smart contract on the Ethereum blockchain to guarantee the atomicity of payment. In Omni-Lytics, a model owner publishes encrypted initial model on the contract, over which the participating data owners compute gradients using their private data, and securely aggregate the gradients through the contract. Finally, the contract reimburses the data owners, and the model owner decrypts the aggregated model update. We implement a working prototype of OmniLytics on Ethereum, and perform extensive experiments to measure its gas cost and execution time under various parameter combinations, demonstrating its high computation and cost efficiency and strong practicality.

show abstract

Section: Security Of Model Updatementioning

confidence: 97%

“…To combat malicious data owners uploading faulty computation results to the contract, we employ the m-Krum algorithm from (Blanchard et al, 2017b) to select the ag-gregation results from a subset of P ⊂ P, which are considered to be close to the expected value with respect to the underlying data distribution.…”

Section: Security Of Model Updatementioning

confidence: 99%

OmniLytics: A Blockchain-based Secure Data Market for Decentralized Machine Learning

Liang

Jiang

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…One major category of these algorithms is called gradient filters [49], or robust gradient aggregation [7,19], which are designed and used mainly with (distributed) gradient descent (abbr. DGD) [79].…”

Section: Gradient Descent With Gradient Filtersmentioning

confidence: 99%

“…Multi-KRUM Multi-Krum [6,7] is a variant of Krum. Instead of selecting one vector, multi-Krum selects m vectors and averages them, where m is a hyperparameter.…”

mentioning

confidence: 99%

A Survey on Fault-tolerance in Distributed Optimization and Machine Learning

Liu

2021

Preprint

View full text Add to dashboard Cite

The robustness of distributed optimization is an emerging field of study, motivated by various applications of distributed optimization including distributed machine learning, distributed sensing, and swarm robotics. With the rapid expansion of the scale of distributed systems, resilient distributed algorithms for optimization are needed, in order to mitigate system failures, communication issues, or even malicious attacks. This survey investigates the current state of fault-tolerance research in distributed optimization, and aims to provide an overview of the existing studies on both fault-tolerant distributed optimization theories and applicable algorithms.

show abstract

“…For example, a gradient descent machine learning algorithm that handles byzantine failures is presented [3,5]. A practical application is Google's Federated Learning where 𝑚 worker machines analyze 𝑁 𝑚 data samples, where 𝑁 is the total number of samples.…”

Section: Related Workmentioning

confidence: 99%

QUDOS: Quorum-Based Cloud-Edge Distributed DNNs for Security Enhanced Industry 4.0

Wallis¹,

Reich²,

Varghese³

et al. 2021

Preprint

View full text Add to dashboard Cite

Distributed machine learning algorithms that employ Deep Neural Networks (DNNs) are widely used in Industry 4.0 applications, such as smart manufacturing. The layers of a DNN can be mapped onto different nodes located in the cloud, edge and shop floor for preserving privacy. The quality of the data that is fed into and processed through the DNN is of utmost importance for critical tasks, such as inspection and quality control. Distributed Data Validation Networks (DDVNs) are used to validate the quality of the data. However, they are prone to single points of failure when an attack occurs. This paper proposes QUDOS, an approach that enhances the security of a distributed DNN that is supported by DDVNs using quorums. The proposed approach allows individual nodes that are corrupted due to an attack to be detected or excluded when the DNN produces an output. Metrics such as corruption factor and success probability of an attack are considered for evaluating the security aspects of DNNs. A simulation study demonstrates that if the number of corrupted nodes is less than a given threshold for decision-making in a quorum, the QUDOS approach always prevents attacks. Furthermore, the study shows that increasing the size of the quorum has a better impact on security than increasing the number of layers. One merit of QUDOS is that it enhances the security of DNNs without requiring any modifications to the algorithm and can therefore be applied to other classes of problems.

show abstract

Byzantine-Tolerant Machine Learning

Cited by 15 publications

References 13 publications

OmniLytics: A Blockchain-based Secure Data Market for Decentralized Machine Learning

OmniLytics: A Blockchain-based Secure Data Market for Decentralized Machine Learning

A Survey on Fault-tolerance in Distributed Optimization and Machine Learning

QUDOS: Quorum-Based Cloud-Edge Distributed DNNs for Security Enhanced Industry 4.0

Contact Info

Product

Resources

About