EvolGAN: Evolutionary Generative Adversarial Networks

Baptiste Roziere, Fabien Teytaud, Vlad Hosu, Hanhe Lin, Jeremy Rapin, Mariia Zameshina, Olivier Teytaud

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

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.

Original languageEnglish
Title of host publicationComputer Vision – ACCV 2020
Subtitle of host publication15th Asian Conference on Computer Vision, 2020, Revised Selected Papers
EditorsHiroshi Ishikawa, Cheng-Lin Liu, Tomas Pajdla, Jianbo Shi
Place of PublicationCham
PublisherSpringer Nature Switzerland AG
Pages679-694
Number of pages16
ISBN (Electronic)978-3-030-69538-5
ISBN (Print)978-3-030-69537-8
DOIs
Publication statusPublished - 2021
Event15th Asian Conference on Computer Vision, ACCV 2020 - Virtual, Online
Duration: 30 Nov 20204 Dec 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12625 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th Asian Conference on Computer Vision, ACCV 2020
CityVirtual, Online
Period30/11/204/12/20

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'EvolGAN: Evolutionary Generative Adversarial Networks'. Together they form a unique fingerprint.

Cite this