Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313396675> ?p ?o ?g. }
Showing items 1 to 100 of
100
with 100 items per page.
- W4313396675 endingPage "830" @default.
- W4313396675 startingPage "818" @default.
- W4313396675 abstract "Federated learning (FL) preserves data privacy by exchanging gradients instead of local training data. However, these private data can still be reconstructed from the exchanged gradients. Deep leakage from gradients (DLG) is a classical reconstruction attack that optimizes dummy data to real data by making the corresponding dummy and real gradients as similar as possible. Nevertheless, DLG fails with highly compressed gradients, which are crucial for communication-efficient FL. In this study, we propose an effective data reconstruction attack against highly compressed gradients, called highly compressed gradient leakage attack (HCGLA). In particular, HCGLA is characterized by the following three key techniques: 1) Owing to the unreasonable optimization objective of DLG in compression scenarios, we redesign a plausible objective function, ensuring that compressed dummy gradients are similar to the compressed real gradients. 2) Instead of simply initializing dummy data through random noise, as in DLG, we design a novel dummy data initialization method, Init-Generation, to compensate for information loss caused by gradient compression. 3) To further enhance reconstruction quality, we train an ad hoc denoising model using the methods of “first optimizing, next filtering, and then reoptimizing”. Extensive experiments on various benchmark data sets and mainstream models show that HCGLA is an effective reconstruction attack even against highly compressed gradients of 0.1%, whereas state-of-the-art attacks can only support 70% compression, thereby achieving a 700-fold improvement." @default.
- W4313396675 created "2023-01-06" @default.
- W4313396675 creator A5003706482 @default.
- W4313396675 creator A5034762138 @default.
- W4313396675 creator A5048390686 @default.
- W4313396675 creator A5064450266 @default.
- W4313396675 date "2023-01-01" @default.
- W4313396675 modified "2023-10-14" @default.
- W4313396675 title "Using Highly Compressed Gradients in Federated Learning for Data Reconstruction Attacks" @default.
- W4313396675 cites W1834627138 @default.
- W4313396675 cites W1902237438 @default.
- W4313396675 cites W2051434435 @default.
- W4313396675 cites W2194775991 @default.
- W4313396675 cites W2804591381 @default.
- W4313396675 cites W2887995258 @default.
- W4313396675 cites W2897830718 @default.
- W4313396675 cites W2912213068 @default.
- W4313396675 cites W2919234133 @default.
- W4313396675 cites W2921560693 @default.
- W4313396675 cites W2963248893 @default.
- W4313396675 cites W2963456518 @default.
- W4313396675 cites W2963540401 @default.
- W4313396675 cites W2964162474 @default.
- W4313396675 cites W2990183786 @default.
- W4313396675 cites W3016560828 @default.
- W4313396675 cites W3024346083 @default.
- W4313396675 cites W3087391814 @default.
- W4313396675 cites W3159069780 @default.
- W4313396675 cites W3175192640 @default.
- W4313396675 cites W4229083957 @default.
- W4313396675 cites W4285722492 @default.
- W4313396675 doi "https://doi.org/10.1109/tifs.2022.3227761" @default.
- W4313396675 hasPublicationYear "2023" @default.
- W4313396675 type Work @default.
- W4313396675 citedByCount "1" @default.
- W4313396675 countsByYear W43133966752023 @default.
- W4313396675 crossrefType "journal-article" @default.
- W4313396675 hasAuthorship W4313396675A5003706482 @default.
- W4313396675 hasAuthorship W4313396675A5034762138 @default.
- W4313396675 hasAuthorship W4313396675A5048390686 @default.
- W4313396675 hasAuthorship W4313396675A5064450266 @default.
- W4313396675 hasConcept C114466953 @default.
- W4313396675 hasConcept C115961682 @default.
- W4313396675 hasConcept C124851039 @default.
- W4313396675 hasConcept C127413603 @default.
- W4313396675 hasConcept C13280743 @default.
- W4313396675 hasConcept C154945302 @default.
- W4313396675 hasConcept C163294075 @default.
- W4313396675 hasConcept C171146098 @default.
- W4313396675 hasConcept C185798385 @default.
- W4313396675 hasConcept C199360897 @default.
- W4313396675 hasConcept C205649164 @default.
- W4313396675 hasConcept C206688291 @default.
- W4313396675 hasConcept C25797200 @default.
- W4313396675 hasConcept C41008148 @default.
- W4313396675 hasConcept C50644808 @default.
- W4313396675 hasConcept C511840579 @default.
- W4313396675 hasConcept C78548338 @default.
- W4313396675 hasConcept C99498987 @default.
- W4313396675 hasConceptScore W4313396675C114466953 @default.
- W4313396675 hasConceptScore W4313396675C115961682 @default.
- W4313396675 hasConceptScore W4313396675C124851039 @default.
- W4313396675 hasConceptScore W4313396675C127413603 @default.
- W4313396675 hasConceptScore W4313396675C13280743 @default.
- W4313396675 hasConceptScore W4313396675C154945302 @default.
- W4313396675 hasConceptScore W4313396675C163294075 @default.
- W4313396675 hasConceptScore W4313396675C171146098 @default.
- W4313396675 hasConceptScore W4313396675C185798385 @default.
- W4313396675 hasConceptScore W4313396675C199360897 @default.
- W4313396675 hasConceptScore W4313396675C205649164 @default.
- W4313396675 hasConceptScore W4313396675C206688291 @default.
- W4313396675 hasConceptScore W4313396675C25797200 @default.
- W4313396675 hasConceptScore W4313396675C41008148 @default.
- W4313396675 hasConceptScore W4313396675C50644808 @default.
- W4313396675 hasConceptScore W4313396675C511840579 @default.
- W4313396675 hasConceptScore W4313396675C78548338 @default.
- W4313396675 hasConceptScore W4313396675C99498987 @default.
- W4313396675 hasFunder F4320316084 @default.
- W4313396675 hasFunder F4320321001 @default.
- W4313396675 hasFunder F4320333335 @default.
- W4313396675 hasFunder F4320335777 @default.
- W4313396675 hasLocation W43133966751 @default.
- W4313396675 hasOpenAccess W4313396675 @default.
- W4313396675 hasPrimaryLocation W43133966751 @default.
- W4313396675 hasRelatedWork W2104750904 @default.
- W4313396675 hasRelatedWork W2116961404 @default.
- W4313396675 hasRelatedWork W2352295696 @default.
- W4313396675 hasRelatedWork W2781597143 @default.
- W4313396675 hasRelatedWork W2888077385 @default.
- W4313396675 hasRelatedWork W2888221816 @default.
- W4313396675 hasRelatedWork W2945877655 @default.
- W4313396675 hasRelatedWork W2996871558 @default.
- W4313396675 hasRelatedWork W3038746171 @default.
- W4313396675 hasRelatedWork W4242031573 @default.
- W4313396675 hasVolume "18" @default.
- W4313396675 isParatext "false" @default.
- W4313396675 isRetracted "false" @default.
- W4313396675 workType "article" @default.