Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386908172> ?p ?o ?g. }
Showing items 1 to 67 of
67
with 100 items per page.
- W4386908172 abstract "While Federated learning (FL) is attractive for pulling privacy-preserving distributed training data, the credibility of participating clients and non-inspectable data pose new security threats, of which poisoning attacks are particularly rampant and hard to defend without compromising privacy, performance or other desirable properties of FL. To tackle this problem, we propose a self-purified FL (SPFL) method that enables benign clients to exploit trusted historical features of locally purified model to supervise the training of aggregated model in each iteration. The purification is performed by an attention-guided self-knowledge distillation where the teacher and student models are optimized locally for task loss, distillation loss and attention-based loss simultaneously. SPFL imposes no restriction on the communication protocol and aggregator at the server. It can work in tandem with any existing secure aggregation algorithms and protocols for augmented security and privacy guarantee. We experimentally demonstrate that SPFL outperforms state-of-the-art FL defenses against various poisoning attacks. The attack success rate of SPFL trained model is at most 3$%$ above that of a clean model, even if the poisoning attack is launched in every iteration with all but one malicious clients in the system. Meantime, it improves the model quality on normal inputs compared to FedAvg, either under attack or in the absence of an attack." @default.
- W4386908172 created "2023-09-21" @default.
- W4386908172 creator A5003261759 @default.
- W4386908172 creator A5017317271 @default.
- W4386908172 creator A5023380073 @default.
- W4386908172 creator A5029335324 @default.
- W4386908172 creator A5048101049 @default.
- W4386908172 creator A5078814357 @default.
- W4386908172 date "2023-09-19" @default.
- W4386908172 modified "2023-10-16" @default.
- W4386908172 title "SPFL: A Self-purified Federated Learning Method Against Poisoning Attacks" @default.
- W4386908172 doi "https://doi.org/10.48550/arxiv.2309.10607" @default.
- W4386908172 hasPublicationYear "2023" @default.
- W4386908172 type Work @default.
- W4386908172 citedByCount "0" @default.
- W4386908172 crossrefType "posted-content" @default.
- W4386908172 hasAuthorship W4386908172A5003261759 @default.
- W4386908172 hasAuthorship W4386908172A5017317271 @default.
- W4386908172 hasAuthorship W4386908172A5023380073 @default.
- W4386908172 hasAuthorship W4386908172A5029335324 @default.
- W4386908172 hasAuthorship W4386908172A5048101049 @default.
- W4386908172 hasAuthorship W4386908172A5078814357 @default.
- W4386908172 hasBestOaLocation W43869081721 @default.
- W4386908172 hasConcept C136764020 @default.
- W4386908172 hasConcept C142724271 @default.
- W4386908172 hasConcept C154945302 @default.
- W4386908172 hasConcept C165696696 @default.
- W4386908172 hasConcept C17744445 @default.
- W4386908172 hasConcept C180505990 @default.
- W4386908172 hasConcept C199539241 @default.
- W4386908172 hasConcept C204787440 @default.
- W4386908172 hasConcept C2780224610 @default.
- W4386908172 hasConcept C2780385302 @default.
- W4386908172 hasConcept C2992525071 @default.
- W4386908172 hasConcept C38652104 @default.
- W4386908172 hasConcept C41008148 @default.
- W4386908172 hasConcept C71924100 @default.
- W4386908172 hasConceptScore W4386908172C136764020 @default.
- W4386908172 hasConceptScore W4386908172C142724271 @default.
- W4386908172 hasConceptScore W4386908172C154945302 @default.
- W4386908172 hasConceptScore W4386908172C165696696 @default.
- W4386908172 hasConceptScore W4386908172C17744445 @default.
- W4386908172 hasConceptScore W4386908172C180505990 @default.
- W4386908172 hasConceptScore W4386908172C199539241 @default.
- W4386908172 hasConceptScore W4386908172C204787440 @default.
- W4386908172 hasConceptScore W4386908172C2780224610 @default.
- W4386908172 hasConceptScore W4386908172C2780385302 @default.
- W4386908172 hasConceptScore W4386908172C2992525071 @default.
- W4386908172 hasConceptScore W4386908172C38652104 @default.
- W4386908172 hasConceptScore W4386908172C41008148 @default.
- W4386908172 hasConceptScore W4386908172C71924100 @default.
- W4386908172 hasLocation W43869081721 @default.
- W4386908172 hasOpenAccess W4386908172 @default.
- W4386908172 hasPrimaryLocation W43869081721 @default.
- W4386908172 hasRelatedWork W1547639832 @default.
- W4386908172 hasRelatedWork W2331043530 @default.
- W4386908172 hasRelatedWork W2374725260 @default.
- W4386908172 hasRelatedWork W2393933887 @default.
- W4386908172 hasRelatedWork W2779961139 @default.
- W4386908172 hasRelatedWork W2997512100 @default.
- W4386908172 hasRelatedWork W3006507989 @default.
- W4386908172 hasRelatedWork W3048799479 @default.
- W4386908172 hasRelatedWork W4280503768 @default.
- W4386908172 hasRelatedWork W4280593074 @default.
- W4386908172 isParatext "false" @default.
- W4386908172 isRetracted "false" @default.
- W4386908172 workType "article" @default.