Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287636610> ?p ?o ?g. }
Showing items 1 to 62 of
62
with 100 items per page.
- W4287636610 abstract "In federated learning, machine learning and deep learning models are trained globally on distributed devices. The state-of-the-art privacy-preserving technique in the context of federated learning is user-level differential privacy. However, such a mechanism is vulnerable to some specific model poisoning attacks such as Sybil attacks. A malicious adversary could create multiple fake clients or collude compromised devices in Sybil attacks to mount direct model updates manipulation. Recent works on novel defense against model poisoning attacks are difficult to detect Sybil attacks when differential privacy is utilized, as it masks clients' model updates with perturbation. In this work, we implement the first Sybil attacks on differential privacy based federated learning architectures and show their impacts on model convergence. We randomly compromise some clients by manipulating different noise levels reflected by the local privacy budget epsilon of differential privacy on the local model updates of these Sybil clients such that the global model convergence rates decrease or even leads to divergence. We apply our attacks to two recent aggregation defense mechanisms, called Krum and Trimmed Mean. Our evaluation results on the MNIST and CIFAR-10 datasets show that our attacks effectively slow down the convergence of the global models. We then propose a method to keep monitoring the average loss of all participants in each round for convergence anomaly detection and defend our Sybil attacks based on the prediction cost reported from each client. Our empirical study demonstrates that our defense approach effectively mitigates the impact of our Sybil attacks on model convergence." @default.
- W4287636610 created "2022-07-25" @default.
- W4287636610 creator A5023214008 @default.
- W4287636610 creator A5029669009 @default.
- W4287636610 creator A5030208871 @default.
- W4287636610 creator A5033033828 @default.
- W4287636610 date "2020-10-20" @default.
- W4287636610 modified "2023-10-18" @default.
- W4287636610 title "Mitigating Sybil Attacks on Differential Privacy based Federated Learning" @default.
- W4287636610 hasPublicationYear "2020" @default.
- W4287636610 type Work @default.
- W4287636610 citedByCount "0" @default.
- W4287636610 crossrefType "posted-content" @default.
- W4287636610 hasAuthorship W4287636610A5023214008 @default.
- W4287636610 hasAuthorship W4287636610A5029669009 @default.
- W4287636610 hasAuthorship W4287636610A5030208871 @default.
- W4287636610 hasAuthorship W4287636610A5033033828 @default.
- W4287636610 hasBestOaLocation W42876366101 @default.
- W4287636610 hasConcept C119857082 @default.
- W4287636610 hasConcept C124101348 @default.
- W4287636610 hasConcept C127413603 @default.
- W4287636610 hasConcept C146978453 @default.
- W4287636610 hasConcept C154945302 @default.
- W4287636610 hasConcept C162324750 @default.
- W4287636610 hasConcept C23130292 @default.
- W4287636610 hasConcept C2777303404 @default.
- W4287636610 hasConcept C2992525071 @default.
- W4287636610 hasConcept C38652104 @default.
- W4287636610 hasConcept C41008148 @default.
- W4287636610 hasConcept C41065033 @default.
- W4287636610 hasConcept C50522688 @default.
- W4287636610 hasConcept C93226319 @default.
- W4287636610 hasConceptScore W4287636610C119857082 @default.
- W4287636610 hasConceptScore W4287636610C124101348 @default.
- W4287636610 hasConceptScore W4287636610C127413603 @default.
- W4287636610 hasConceptScore W4287636610C146978453 @default.
- W4287636610 hasConceptScore W4287636610C154945302 @default.
- W4287636610 hasConceptScore W4287636610C162324750 @default.
- W4287636610 hasConceptScore W4287636610C23130292 @default.
- W4287636610 hasConceptScore W4287636610C2777303404 @default.
- W4287636610 hasConceptScore W4287636610C2992525071 @default.
- W4287636610 hasConceptScore W4287636610C38652104 @default.
- W4287636610 hasConceptScore W4287636610C41008148 @default.
- W4287636610 hasConceptScore W4287636610C41065033 @default.
- W4287636610 hasConceptScore W4287636610C50522688 @default.
- W4287636610 hasConceptScore W4287636610C93226319 @default.
- W4287636610 hasLocation W42876366101 @default.
- W4287636610 hasOpenAccess W4287636610 @default.
- W4287636610 hasPrimaryLocation W42876366101 @default.
- W4287636610 hasRelatedWork W13671444 @default.
- W4287636610 hasRelatedWork W14398808 @default.
- W4287636610 hasRelatedWork W2183821 @default.
- W4287636610 hasRelatedWork W3861039 @default.
- W4287636610 hasRelatedWork W601863 @default.
- W4287636610 hasRelatedWork W6745555 @default.
- W4287636610 hasRelatedWork W6908809 @default.
- W4287636610 hasRelatedWork W7859117 @default.
- W4287636610 hasRelatedWork W8224326 @default.
- W4287636610 hasRelatedWork W9657784 @default.
- W4287636610 isParatext "false" @default.
- W4287636610 isRetracted "false" @default.
- W4287636610 workType "article" @default.