Fréchet AutoEncoder Distance: A new approach for evaluation of Generative Adversarial Networks

Página do item simplificado
dc.contributor.advisorOrcidhttps://orcid.org/0000-0001-5566-1963
dc.contributor.authorBUZUTI, L. F.
dc.contributor.authorThomaz C. E.
dc.date.accessioned2023-08-01T06:02:49Z
dc.date.available2023-08-01T06:02:49Z
dc.date.issued2023-10-05
dc.description.abstract© 2023 Elsevier Inc.Evaluation measures of Generative Adversarial Networks (GANs) have been an active area of research and, currently, there are several measures to evaluate them. The most used GANs evaluation measure is the Fréchet Inception Distance (FID). Measures such as FID are known as model-agnostic methods, where the generator is used as a black box to sample the generated images. Like other measures of model-agnostic, FID uses a deep supervised model for mapping real and generated samples to a feature space. We proposed an approach here with a deep unsupervised model, the Vector Quantised-Variational Autoencoder (VQ-VAE), for estimating the mean and the covariance matrix of the Fréchet Distance and named it Fréchet AutoEncoder Distance (FAED). Our experimental results highlighted that the feature space of the VQ-VAE describes a clustering domain-specific representation more intuitive and visually plausible than the Inception network used by the benchmark FID.
dc.description.volume235
dc.identifier.citationBUZUTI, L. F.; Thomaz C. E. Fréchet autoencoder distance: a new approach for evaluation of generative adversarial networks. Computer Vision and Image Understanding, v. 235, oct. 2023.
dc.identifier.doi10.1016/j.cviu.2023.103768
dc.identifier.issn1077-3142
dc.identifier.urihttps://repositorio.fei.edu.br/handle/FEI/4864
dc.relation.ispartofComputer Vision and Image Understanding
dc.rightsAcesso Restrito
dc.subject.otherlanguageAutoencoder
dc.subject.otherlanguageEvaluation
dc.subject.otherlanguageFeature extraction
dc.subject.otherlanguageFréchet distance
dc.subject.otherlanguageGenerative adversarial networks
dc.subject.otherlanguageMeasure
dc.titleFréchet AutoEncoder Distance: A new approach for evaluation of Generative Adversarial Networks
dc.typeArtigo
fei.scopus.citations1
fei.scopus.eid2-s2.0-85164703000
fei.scopus.subjectActive area
fei.scopus.subjectAuto encoders
fei.scopus.subjectEvaluation
fei.scopus.subjectEvaluation measures
fei.scopus.subjectFeature space
fei.scopus.subjectFeatures extraction
fei.scopus.subjectFrechet
fei.scopus.subjectFrechet distance
fei.scopus.subjectMeasure
fei.scopus.subjectNew approaches
fei.scopus.updated2024-07-01
fei.scopus.urlhttps://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85164703000&origin=inward
Arquivos
Coleções
Artigos