Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313886738> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W4313886738 endingPage "3934" @default.
- W4313886738 startingPage "3924" @default.
- W4313886738 abstract "Label noise is a ubiquitous issue in GANs, which degrades the generalization ability of the discriminator and usually leads to instability when training GANs. This issue stems from both real data and generated data. Previous works either only consider one of these two sources, or are not robust enough to noisy labels. In this paper, we revisit spectral normalization in robust learning with noisy labels. Based on its pros and cons, we propose to combine spectral normalization and weight decay to regularize the discriminator, which enjoys a more robust training process. To extend to conditional GANs, we propose to balance the relative importance of marginal matching and conditional matching in the projection discriminator. The proposed Enhanced Spectral Normalization for Generative Adversarial Networks (ESNGAN) can be easily integrated into various existing GANs frameworks without excessive additional cost. The effectiveness of the proposed method is validated on the CIFAR10, LSUN Church, CelebA, and ImageNet datasets, including the unconditional image generation task and the class-conditional image generation task. We also show that the proposed method can further improve the performance of the high-resolution image generation task." @default.
- W4313886738 created "2023-01-10" @default.
- W4313886738 creator A5016322200 @default.
- W4313886738 creator A5049158294 @default.
- W4313886738 creator A5050309466 @default.
- W4313886738 creator A5056366783 @default.
- W4313886738 creator A5057436575 @default.
- W4313886738 date "2023-08-01" @default.
- W4313886738 modified "2023-09-26" @default.
- W4313886738 title "Mitigating Label Noise in GANs via Enhanced Spectral Normalization" @default.
- W4313886738 cites W1834627138 @default.
- W4313886738 cites W2108598243 @default.
- W4313886738 cites W2194775991 @default.
- W4313886738 cites W2914733350 @default.
- W4313886738 cites W2953327099 @default.
- W4313886738 cites W2962770929 @default.
- W4313886738 cites W2962793481 @default.
- W4313886738 cites W2963073614 @default.
- W4313886738 cites W2963470893 @default.
- W4313886738 cites W2964292098 @default.
- W4313886738 cites W2964343197 @default.
- W4313886738 cites W2981873476 @default.
- W4313886738 cites W2986510250 @default.
- W4313886738 cites W2990251202 @default.
- W4313886738 cites W2990440529 @default.
- W4313886738 cites W3035235751 @default.
- W4313886738 cites W3108650314 @default.
- W4313886738 cites W3117074306 @default.
- W4313886738 cites W3118813946 @default.
- W4313886738 cites W3166279462 @default.
- W4313886738 cites W3173275980 @default.
- W4313886738 cites W3174568369 @default.
- W4313886738 cites W3193773061 @default.
- W4313886738 cites W4214585162 @default.
- W4313886738 doi "https://doi.org/10.1109/tcsvt.2023.3235410" @default.
- W4313886738 hasPublicationYear "2023" @default.
- W4313886738 type Work @default.
- W4313886738 citedByCount "0" @default.
- W4313886738 crossrefType "journal-article" @default.
- W4313886738 hasAuthorship W4313886738A5016322200 @default.
- W4313886738 hasAuthorship W4313886738A5049158294 @default.
- W4313886738 hasAuthorship W4313886738A5050309466 @default.
- W4313886738 hasAuthorship W4313886738A5056366783 @default.
- W4313886738 hasAuthorship W4313886738A5057436575 @default.
- W4313886738 hasConcept C115961682 @default.
- W4313886738 hasConcept C136886441 @default.
- W4313886738 hasConcept C144024400 @default.
- W4313886738 hasConcept C153180895 @default.
- W4313886738 hasConcept C154945302 @default.
- W4313886738 hasConcept C19165224 @default.
- W4313886738 hasConcept C2779803651 @default.
- W4313886738 hasConcept C41008148 @default.
- W4313886738 hasConcept C76155785 @default.
- W4313886738 hasConcept C94915269 @default.
- W4313886738 hasConcept C99498987 @default.
- W4313886738 hasConceptScore W4313886738C115961682 @default.
- W4313886738 hasConceptScore W4313886738C136886441 @default.
- W4313886738 hasConceptScore W4313886738C144024400 @default.
- W4313886738 hasConceptScore W4313886738C153180895 @default.
- W4313886738 hasConceptScore W4313886738C154945302 @default.
- W4313886738 hasConceptScore W4313886738C19165224 @default.
- W4313886738 hasConceptScore W4313886738C2779803651 @default.
- W4313886738 hasConceptScore W4313886738C41008148 @default.
- W4313886738 hasConceptScore W4313886738C76155785 @default.
- W4313886738 hasConceptScore W4313886738C94915269 @default.
- W4313886738 hasConceptScore W4313886738C99498987 @default.
- W4313886738 hasFunder F4320321001 @default.
- W4313886738 hasFunder F4320329791 @default.
- W4313886738 hasIssue "8" @default.
- W4313886738 hasLocation W43138867381 @default.
- W4313886738 hasOpenAccess W4313886738 @default.
- W4313886738 hasPrimaryLocation W43138867381 @default.
- W4313886738 hasRelatedWork W1991269640 @default.
- W4313886738 hasRelatedWork W2016839265 @default.
- W4313886738 hasRelatedWork W2785678896 @default.
- W4313886738 hasRelatedWork W2963836885 @default.
- W4313886738 hasRelatedWork W2965098853 @default.
- W4313886738 hasRelatedWork W3198909301 @default.
- W4313886738 hasRelatedWork W3203390936 @default.
- W4313886738 hasRelatedWork W4280544492 @default.
- W4313886738 hasRelatedWork W4321441197 @default.
- W4313886738 hasRelatedWork W2508457823 @default.
- W4313886738 hasVolume "33" @default.
- W4313886738 isParatext "false" @default.
- W4313886738 isRetracted "false" @default.
- W4313886738 workType "article" @default.