%0 Conference Paper %1 Roziere_2020_ACCV %A Roziere, Baptiste %A Teytaud, Fabien %A Hosu, Vlad %A Lin, Hanhe %A Rapin, Jeremy %A Zameshina, Mariia %A Teytaud, Olivier %B Computer Vision -- ACCV 2020 %C Cham %D 2021 %I Springer International Publishing %K 2021 a05 from:christinawarren sfbtrr161 %P 679--694 %R 10.1007/978-3-030-69538-5_41 %T EvolGAN: Evolutionary Generative Adversarial Networks %U https://openaccess.thecvf.com/content/ACCV2020/html/Roziere_EvolGAN_Evolutionary_Generative_Adversarial_Networks_ACCV_2020_paper.html %X We propose to use a quality estimator and evolutionary methods to search the latent space of generative adversarial networks trained on small, difficult datasets, or both. The new method leads to the generation of significantly higher quality images while preserving the original generator’s diversity. Human raters preferred an image from the new version with frequency 83.7% for Cats, 74% for FashionGen, 70.4% for Horses, and 69.2% for Artworks - minor improvements for the already excellent GANs for faces. This approach applies to any quality scorer and GAN generator.

@inproceedings{Roziere_2020_ACCV, abstract = {We propose to use a quality estimator and evolutionary methods to search the latent space of generative adversarial networks trained on small, difficult datasets, or both. The new method leads to the generation of significantly higher quality images while preserving the original generator’s diversity. Human raters preferred an image from the new version with frequency 83.7% for Cats, 74% for FashionGen, 70.4% for Horses, and 69.2% for Artworks - minor improvements for the already excellent GANs for faces. This approach applies to any quality scorer and GAN generator.}, added-at = {2021-06-23T09:48:36.000+0200}, address = {Cham}, author = {Roziere, Baptiste and Teytaud, Fabien and Hosu, Vlad and Lin, Hanhe and Rapin, Jeremy and Zameshina, Mariia and Teytaud, Olivier}, biburl = {https://puma.ub.uni-stuttgart.de/bibtex/2d435ceeb13e771bf463189f742db2f81/sfbtrr161}, booktitle = {Computer Vision -- ACCV 2020}, doi = {10.1007/978-3-030-69538-5_41}, interhash = {70e8c5a3c4e3c9ab758aebdf93bd6062}, intrahash = {d435ceeb13e771bf463189f742db2f81}, keywords = {2021 a05 from:christinawarren sfbtrr161}, month = {11}, pages = {679--694}, publisher = {Springer International Publishing}, timestamp = {2021-07-23T07:54:32.000+0200}, title = {EvolGAN: Evolutionary Generative Adversarial Networks}, url = {https://openaccess.thecvf.com/content/ACCV2020/html/Roziere_EvolGAN_Evolutionary_Generative_Adversarial_Networks_ACCV_2020_paper.html}, year = 2021 }