Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386066557> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4386066557 abstract "Privacy-preserving image generation has been important for segments such as medical domains that have sensitive and limited data. The benefits of guaranteed privacy come at the costs of generated images' quality and utility due to the privacy budget constraints. The utility is currently measured by the gen2real accuracy (g2r%), i.e., the accuracy on real data of a downstream classifier trained using generated data. However, apart from this standard utility, we identify the “reversed utility” as another crucial aspect, which computes the accuracy on generated data of a classifier trained using real data, dubbed as real2gen accuracy (r2g%). Jointly considering these two views of utility, the standard and the reversed, could help the generation model better improve transferability between fake and real data. Therefore, we propose a novel private image generation method that incorporates a dual-purpose auxiliary classifier, which alternates between learning from real data and fake data, into the training of differentially private GANs. Additionally, our deliberate training strategies such as sequential training contributes to accelerating the generator's convergence and further boosting the performance upon exhausting the privacy budget. Our results achieve new state-of-the-arts over all metrics on three benchmarks: MNIST, Fashion-MNIST, and CelebA." @default.
- W4386066557 created "2023-08-23" @default.
- W4386066557 creator A5001529504 @default.
- W4386066557 creator A5016972157 @default.
- W4386066557 creator A5029813674 @default.
- W4386066557 creator A5063253432 @default.
- W4386066557 creator A5082256444 @default.
- W4386066557 date "2023-06-01" @default.
- W4386066557 modified "2023-10-06" @default.
- W4386066557 title "Private Image Generation with Dual-Purpose Auxiliary Classifier" @default.
- W4386066557 cites W2051267297 @default.
- W4386066557 cites W2112796928 @default.
- W4386066557 cites W2167372639 @default.
- W4386066557 cites W2473418344 @default.
- W4386066557 cites W2535690855 @default.
- W4386066557 cites W2566079294 @default.
- W4386066557 cites W2587706859 @default.
- W4386066557 cites W2794022343 @default.
- W4386066557 cites W2887995258 @default.
- W4386066557 cites W2910022251 @default.
- W4386066557 cites W2911915417 @default.
- W4386066557 cites W2961396908 @default.
- W4386066557 cites W2963073614 @default.
- W4386066557 cites W2963470893 @default.
- W4386066557 cites W2967880504 @default.
- W4386066557 cites W2984306354 @default.
- W4386066557 cites W3192357745 @default.
- W4386066557 cites W4200218970 @default.
- W4386066557 doi "https://doi.org/10.1109/cvpr52729.2023.01950" @default.
- W4386066557 hasPublicationYear "2023" @default.
- W4386066557 type Work @default.
- W4386066557 citedByCount "0" @default.
- W4386066557 crossrefType "proceedings-article" @default.
- W4386066557 hasAuthorship W4386066557A5001529504 @default.
- W4386066557 hasAuthorship W4386066557A5016972157 @default.
- W4386066557 hasAuthorship W4386066557A5029813674 @default.
- W4386066557 hasAuthorship W4386066557A5063253432 @default.
- W4386066557 hasAuthorship W4386066557A5082256444 @default.
- W4386066557 hasConcept C108583219 @default.
- W4386066557 hasConcept C119857082 @default.
- W4386066557 hasConcept C124101348 @default.
- W4386066557 hasConcept C153180895 @default.
- W4386066557 hasConcept C154945302 @default.
- W4386066557 hasConcept C160920958 @default.
- W4386066557 hasConcept C190502265 @default.
- W4386066557 hasConcept C41008148 @default.
- W4386066557 hasConcept C51632099 @default.
- W4386066557 hasConcept C95623464 @default.
- W4386066557 hasConceptScore W4386066557C108583219 @default.
- W4386066557 hasConceptScore W4386066557C119857082 @default.
- W4386066557 hasConceptScore W4386066557C124101348 @default.
- W4386066557 hasConceptScore W4386066557C153180895 @default.
- W4386066557 hasConceptScore W4386066557C154945302 @default.
- W4386066557 hasConceptScore W4386066557C160920958 @default.
- W4386066557 hasConceptScore W4386066557C190502265 @default.
- W4386066557 hasConceptScore W4386066557C41008148 @default.
- W4386066557 hasConceptScore W4386066557C51632099 @default.
- W4386066557 hasConceptScore W4386066557C95623464 @default.
- W4386066557 hasFunder F4320334704 @default.
- W4386066557 hasLocation W43860665571 @default.
- W4386066557 hasOpenAccess W4386066557 @default.
- W4386066557 hasPrimaryLocation W43860665571 @default.
- W4386066557 hasRelatedWork W2088804257 @default.
- W4386066557 hasRelatedWork W2167582322 @default.
- W4386066557 hasRelatedWork W2563096758 @default.
- W4386066557 hasRelatedWork W2742991909 @default.
- W4386066557 hasRelatedWork W2810865670 @default.
- W4386066557 hasRelatedWork W2972035100 @default.
- W4386066557 hasRelatedWork W4243326899 @default.
- W4386066557 hasRelatedWork W4386053843 @default.
- W4386066557 hasRelatedWork W4386066557 @default.
- W4386066557 hasRelatedWork W3158004940 @default.
- W4386066557 isParatext "false" @default.
- W4386066557 isRetracted "false" @default.
- W4386066557 workType "article" @default.