Matches in SemOpenAlex for { <https://semopenalex.org/work/W2911109055> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W2911109055 abstract "State of the art deep generative networks are capable of producing images with such incredible realism that they can be suspected of memorizing training images. It is why it is not uncommon to include visualizations of training set nearest neighbors, to suggest generated images are not simply memorized. We demonstrate this is not sufficient and motivates the need to study memorization/overfitting of deep generators with more scrutiny. This paper addresses this question by i) showing how simple losses are highly effective at reconstructing images for deep generators ii) analyzing the statistics of reconstruction errors when reconstructing training and validation images, which is the standard way to analyze overfitting in machine learning. Using this methodology, this paper shows that overfitting is not detectable in the pure GAN models proposed in the literature, in contrast with those using hybrid adversarial losses, which are amongst the most widely applied generative methods. The paper also shows that standard GAN evaluation metrics fail to capture memorization for some deep generators. Finally, the paper also shows how off-the-shelf GAN generators can be successfully applied to face inpainting and face super-resolution using the proposed reconstruction method, without hybrid adversarial losses." @default.
- W2911109055 created "2019-01-25" @default.
- W2911109055 creator A5001375778 @default.
- W2911109055 creator A5030910160 @default.
- W2911109055 creator A5072699551 @default.
- W2911109055 creator A5082779714 @default.
- W2911109055 date "2019-06-16" @default.
- W2911109055 modified "2023-09-23" @default.
- W2911109055 title "Detecting Overfitting of Deep Generative Networks via Latent Recovery" @default.
- W2911109055 cites W1915485278 @default.
- W2911109055 cites W1932198206 @default.
- W2911109055 cites W1976101156 @default.
- W2911109055 cites W2051267297 @default.
- W2911109055 cites W2099471712 @default.
- W2911109055 cites W2173520492 @default.
- W2911109055 cites W2331128040 @default.
- W2911109055 cites W2535690855 @default.
- W2911109055 cites W2554314924 @default.
- W2911109055 cites W2593729559 @default.
- W2911109055 cites W2595294663 @default.
- W2911109055 cites W2605195953 @default.
- W2911109055 cites W2611104282 @default.
- W2911109055 cites W2738588019 @default.
- W2911109055 cites W2785838823 @default.
- W2911109055 cites W2792263949 @default.
- W2911109055 cites W2805984778 @default.
- W2911109055 cites W2893749619 @default.
- W2911109055 cites W2962760235 @default.
- W2911109055 cites W2962785568 @default.
- W2911109055 cites W2962793481 @default.
- W2911109055 cites W2962808998 @default.
- W2911109055 cites W2963185411 @default.
- W2911109055 cites W2963186101 @default.
- W2911109055 cites W2963373786 @default.
- W2911109055 cites W2963376432 @default.
- W2911109055 cites W2963767194 @default.
- W2911109055 cites W2963917315 @default.
- W2911109055 cites W2963981733 @default.
- W2911109055 cites W2964013315 @default.
- W2911109055 cites W2964153729 @default.
- W2911109055 cites W2964240153 @default.
- W2911109055 cites W3137695714 @default.
- W2911109055 hasPublicationYear "2019" @default.
- W2911109055 type Work @default.
- W2911109055 sameAs 2911109055 @default.
- W2911109055 citedByCount "3" @default.
- W2911109055 countsByYear W29111090552019 @default.
- W2911109055 countsByYear W29111090552021 @default.
- W2911109055 crossrefType "proceedings-article" @default.
- W2911109055 hasAuthorship W2911109055A5001375778 @default.
- W2911109055 hasAuthorship W2911109055A5030910160 @default.
- W2911109055 hasAuthorship W2911109055A5072699551 @default.
- W2911109055 hasAuthorship W2911109055A5082779714 @default.
- W2911109055 hasBestOaLocation W29111090552 @default.
- W2911109055 hasConcept C119857082 @default.
- W2911109055 hasConcept C153180895 @default.
- W2911109055 hasConcept C154945302 @default.
- W2911109055 hasConcept C167966045 @default.
- W2911109055 hasConcept C22019652 @default.
- W2911109055 hasConcept C39890363 @default.
- W2911109055 hasConcept C41008148 @default.
- W2911109055 hasConcept C50644808 @default.
- W2911109055 hasConceptScore W2911109055C119857082 @default.
- W2911109055 hasConceptScore W2911109055C153180895 @default.
- W2911109055 hasConceptScore W2911109055C154945302 @default.
- W2911109055 hasConceptScore W2911109055C167966045 @default.
- W2911109055 hasConceptScore W2911109055C22019652 @default.
- W2911109055 hasConceptScore W2911109055C39890363 @default.
- W2911109055 hasConceptScore W2911109055C41008148 @default.
- W2911109055 hasConceptScore W2911109055C50644808 @default.
- W2911109055 hasLocation W29111090551 @default.
- W2911109055 hasLocation W29111090552 @default.
- W2911109055 hasOpenAccess W2911109055 @default.
- W2911109055 hasPrimaryLocation W29111090551 @default.
- W2911109055 hasRelatedWork W2796067775 @default.
- W2911109055 hasRelatedWork W2883697671 @default.
- W2911109055 hasRelatedWork W2890766902 @default.
- W2911109055 hasRelatedWork W2949895856 @default.
- W2911109055 hasRelatedWork W2985459377 @default.
- W2911109055 hasRelatedWork W2989932438 @default.
- W2911109055 hasRelatedWork W3099765033 @default.
- W2911109055 hasRelatedWork W3129477419 @default.
- W2911109055 hasRelatedWork W3175189414 @default.
- W2911109055 hasRelatedWork W4210794429 @default.
- W2911109055 isParatext "false" @default.
- W2911109055 isRetracted "false" @default.
- W2911109055 magId "2911109055" @default.
- W2911109055 workType "article" @default.