Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204999899> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W3204999899 abstract "Unlike traditional central training, federated learning (FL) improves the performance of the global model by sharing and aggregating local models rather than local data to protect the users' privacy. Although this training approach appears secure, some research has demonstrated that an attacker can still recover private data based on the shared gradient information. This on-the-fly reconstruction attack deserves to be studied in depth because it can occur at any stage of training, whether at the beginning or at the end of model training; no relevant dataset is required and no additional models need to be trained. We break through some unrealistic assumptions and limitations to apply this reconstruction attack in a broader range of scenarios. We propose methods that can reconstruct the training data from shared gradients or weights, corresponding to the FedSGD and FedAvg usage scenarios, respectively. We propose a zero-shot approach to restore labels even if there are duplicate labels in the batch. We study the relationship between the label and image restoration. We find that image restoration fails even if there is only one incorrectly inferred label in the batch; we also find that when batch images have the same label, the corresponding image is restored as a fusion of that class of images. Our approaches are evaluated on classic image benchmarks, including CIFAR-10 and ImageNet. The batch size, image quality, and the adaptability of the label distribution of our approach exceed those of GradInversion, the state-of-the-art." @default.
- W3204999899 created "2021-10-25" @default.
- W3204999899 creator A5014301216 @default.
- W3204999899 creator A5027466503 @default.
- W3204999899 creator A5053780153 @default.
- W3204999899 creator A5067635650 @default.
- W3204999899 creator A5071104283 @default.
- W3204999899 creator A5073054282 @default.
- W3204999899 creator A5082946919 @default.
- W3204999899 date "2021-10-18" @default.
- W3204999899 modified "2023-10-16" @default.
- W3204999899 title "Towards General Deep Leakage in Federated Learning" @default.
- W3204999899 cites W1915485278 @default.
- W3204999899 cites W2051267297 @default.
- W3204999899 cites W2283463896 @default.
- W3204999899 cites W2541884796 @default.
- W3204999899 cites W2930926105 @default.
- W3204999899 cites W2963456518 @default.
- W3204999899 cites W2964162474 @default.
- W3204999899 cites W2970408908 @default.
- W3204999899 cites W3000479830 @default.
- W3204999899 cites W3014541599 @default.
- W3204999899 cites W3048684575 @default.
- W3204999899 cites W3175192640 @default.
- W3204999899 doi "https://doi.org/10.48550/arxiv.2110.09074" @default.
- W3204999899 hasPublicationYear "2021" @default.
- W3204999899 type Work @default.
- W3204999899 sameAs 3204999899 @default.
- W3204999899 citedByCount "0" @default.
- W3204999899 crossrefType "posted-content" @default.
- W3204999899 hasAuthorship W3204999899A5014301216 @default.
- W3204999899 hasAuthorship W3204999899A5027466503 @default.
- W3204999899 hasAuthorship W3204999899A5053780153 @default.
- W3204999899 hasAuthorship W3204999899A5067635650 @default.
- W3204999899 hasAuthorship W3204999899A5071104283 @default.
- W3204999899 hasAuthorship W3204999899A5073054282 @default.
- W3204999899 hasAuthorship W3204999899A5082946919 @default.
- W3204999899 hasBestOaLocation W32049998991 @default.
- W3204999899 hasConcept C115961682 @default.
- W3204999899 hasConcept C119857082 @default.
- W3204999899 hasConcept C124101348 @default.
- W3204999899 hasConcept C139719470 @default.
- W3204999899 hasConcept C154945302 @default.
- W3204999899 hasConcept C159985019 @default.
- W3204999899 hasConcept C162324750 @default.
- W3204999899 hasConcept C177606310 @default.
- W3204999899 hasConcept C18903297 @default.
- W3204999899 hasConcept C192562407 @default.
- W3204999899 hasConcept C204323151 @default.
- W3204999899 hasConcept C2777042071 @default.
- W3204999899 hasConcept C2777212361 @default.
- W3204999899 hasConcept C41008148 @default.
- W3204999899 hasConcept C51632099 @default.
- W3204999899 hasConcept C86803240 @default.
- W3204999899 hasConceptScore W3204999899C115961682 @default.
- W3204999899 hasConceptScore W3204999899C119857082 @default.
- W3204999899 hasConceptScore W3204999899C124101348 @default.
- W3204999899 hasConceptScore W3204999899C139719470 @default.
- W3204999899 hasConceptScore W3204999899C154945302 @default.
- W3204999899 hasConceptScore W3204999899C159985019 @default.
- W3204999899 hasConceptScore W3204999899C162324750 @default.
- W3204999899 hasConceptScore W3204999899C177606310 @default.
- W3204999899 hasConceptScore W3204999899C18903297 @default.
- W3204999899 hasConceptScore W3204999899C192562407 @default.
- W3204999899 hasConceptScore W3204999899C204323151 @default.
- W3204999899 hasConceptScore W3204999899C2777042071 @default.
- W3204999899 hasConceptScore W3204999899C2777212361 @default.
- W3204999899 hasConceptScore W3204999899C41008148 @default.
- W3204999899 hasConceptScore W3204999899C51632099 @default.
- W3204999899 hasConceptScore W3204999899C86803240 @default.
- W3204999899 hasLocation W32049998991 @default.
- W3204999899 hasOpenAccess W3204999899 @default.
- W3204999899 hasPrimaryLocation W32049998991 @default.
- W3204999899 hasRelatedWork W2614183994 @default.
- W3204999899 hasRelatedWork W2953648989 @default.
- W3204999899 hasRelatedWork W3099765033 @default.
- W3204999899 hasRelatedWork W3126776133 @default.
- W3204999899 hasRelatedWork W3133521594 @default.
- W3204999899 hasRelatedWork W3177008965 @default.
- W3204999899 hasRelatedWork W4287234591 @default.
- W3204999899 hasRelatedWork W4295132169 @default.
- W3204999899 hasRelatedWork W4308823300 @default.
- W3204999899 hasRelatedWork W4310034804 @default.
- W3204999899 isParatext "false" @default.
- W3204999899 isRetracted "false" @default.
- W3204999899 magId "3204999899" @default.
- W3204999899 workType "article" @default.