Matches in SemOpenAlex for { <https://semopenalex.org/work/W4229820657> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W4229820657 abstract "Model inversion (MI) attacks are aimed at reconstructing training data from model parameters. Such attacks have triggered increasing concerns about privacy, especially given a growing number of online model repositories. However, existing MI attacks against deep neural networks (DNNs) have large room for performance improvement. We present a novel inversion-specific GAN that can better distill knowledge useful for performing attacks on private models from public data. In particular, we train the discriminator to differentiate not only the real and fake samples but the soft-labels provided by the target model. Moreover, unlike previous work that directly searches for a single data point to represent a target class, we propose to model a private data distribution for each target class. Our experiments show that the combination of these techniques can significantly boost the success rate of the state-of-the-art MI attacks by 150%, and generalize better to a variety of datasets and models. Our code is available at https://github.com/SCccc21/Knowledge-Enriched-DMI." @default.
- W4229820657 created "2022-05-11" @default.
- W4229820657 creator A5030458072 @default.
- W4229820657 creator A5032275274 @default.
- W4229820657 creator A5041294365 @default.
- W4229820657 creator A5082230753 @default.
- W4229820657 date "2021-10-01" @default.
- W4229820657 modified "2023-10-13" @default.
- W4229820657 title "Knowledge-Enriched Distributional Model Inversion Attacks" @default.
- W4229820657 cites W1834627138 @default.
- W4229820657 cites W2024922353 @default.
- W4229820657 cites W2051267297 @default.
- W4229820657 cites W2112796928 @default.
- W4229820657 cites W2194775991 @default.
- W4229820657 cites W2535690855 @default.
- W4229820657 cites W2780958074 @default.
- W4229820657 cites W2808195004 @default.
- W4229820657 cites W2897830718 @default.
- W4229820657 cites W2962770929 @default.
- W4229820657 cites W2962785568 @default.
- W4229820657 cites W2962835266 @default.
- W4229820657 cites W2963303354 @default.
- W4229820657 cites W2963456518 @default.
- W4229820657 cites W3035616549 @default.
- W4229820657 cites W3101156210 @default.
- W4229820657 doi "https://doi.org/10.1109/iccv48922.2021.01587" @default.
- W4229820657 hasPublicationYear "2021" @default.
- W4229820657 type Work @default.
- W4229820657 citedByCount "19" @default.
- W4229820657 countsByYear W42298206572022 @default.
- W4229820657 countsByYear W42298206572023 @default.
- W4229820657 crossrefType "proceedings-article" @default.
- W4229820657 hasAuthorship W4229820657A5030458072 @default.
- W4229820657 hasAuthorship W4229820657A5032275274 @default.
- W4229820657 hasAuthorship W4229820657A5041294365 @default.
- W4229820657 hasAuthorship W4229820657A5082230753 @default.
- W4229820657 hasBestOaLocation W42298206572 @default.
- W4229820657 hasConcept C109007969 @default.
- W4229820657 hasConcept C111919701 @default.
- W4229820657 hasConcept C119857082 @default.
- W4229820657 hasConcept C124101348 @default.
- W4229820657 hasConcept C136197465 @default.
- W4229820657 hasConcept C151730666 @default.
- W4229820657 hasConcept C154945302 @default.
- W4229820657 hasConcept C177264268 @default.
- W4229820657 hasConcept C1893757 @default.
- W4229820657 hasConcept C199360897 @default.
- W4229820657 hasConcept C2776760102 @default.
- W4229820657 hasConcept C2779803651 @default.
- W4229820657 hasConcept C41008148 @default.
- W4229820657 hasConcept C43126263 @default.
- W4229820657 hasConcept C67186912 @default.
- W4229820657 hasConcept C76155785 @default.
- W4229820657 hasConcept C77088390 @default.
- W4229820657 hasConcept C86803240 @default.
- W4229820657 hasConcept C94915269 @default.
- W4229820657 hasConceptScore W4229820657C109007969 @default.
- W4229820657 hasConceptScore W4229820657C111919701 @default.
- W4229820657 hasConceptScore W4229820657C119857082 @default.
- W4229820657 hasConceptScore W4229820657C124101348 @default.
- W4229820657 hasConceptScore W4229820657C136197465 @default.
- W4229820657 hasConceptScore W4229820657C151730666 @default.
- W4229820657 hasConceptScore W4229820657C154945302 @default.
- W4229820657 hasConceptScore W4229820657C177264268 @default.
- W4229820657 hasConceptScore W4229820657C1893757 @default.
- W4229820657 hasConceptScore W4229820657C199360897 @default.
- W4229820657 hasConceptScore W4229820657C2776760102 @default.
- W4229820657 hasConceptScore W4229820657C2779803651 @default.
- W4229820657 hasConceptScore W4229820657C41008148 @default.
- W4229820657 hasConceptScore W4229820657C43126263 @default.
- W4229820657 hasConceptScore W4229820657C67186912 @default.
- W4229820657 hasConceptScore W4229820657C76155785 @default.
- W4229820657 hasConceptScore W4229820657C77088390 @default.
- W4229820657 hasConceptScore W4229820657C86803240 @default.
- W4229820657 hasConceptScore W4229820657C94915269 @default.
- W4229820657 hasLocation W42298206571 @default.
- W4229820657 hasLocation W42298206572 @default.
- W4229820657 hasOpenAccess W4229820657 @default.
- W4229820657 hasPrimaryLocation W42298206571 @default.
- W4229820657 hasRelatedWork W1964938328 @default.
- W4229820657 hasRelatedWork W2070486197 @default.
- W4229820657 hasRelatedWork W2387995142 @default.
- W4229820657 hasRelatedWork W2961085424 @default.
- W4229820657 hasRelatedWork W2999741285 @default.
- W4229820657 hasRelatedWork W3025331699 @default.
- W4229820657 hasRelatedWork W4280544492 @default.
- W4229820657 hasRelatedWork W4287905443 @default.
- W4229820657 hasRelatedWork W4306674287 @default.
- W4229820657 hasRelatedWork W4224009465 @default.
- W4229820657 isParatext "false" @default.
- W4229820657 isRetracted "false" @default.
- W4229820657 workType "article" @default.