Detecting overfitting in GANs with a metric based on the Fourier spectrum

Gamazo Tejero, Ángel Javier. (2020). Detecting overfitting in GANs with a metric based on the Fourier spectrum Master Thesis, Universidad Nacional de Educación a Distancia (España). Escuela Técnica Superior de Ingeniería Informática. Departamento de Inteligencia Artificial

Ficheros (Some files may be inaccessible until you login with your e-spacio credentials)
Nombre Descripción Tipo MIME Size
Gamazo_Tejero_Angel_Javier_TFM.pdf Gamazo_Tejero_Angel_Javier_TFM.pdf application/pdf 7.48MB

Título Detecting overfitting in GANs with a metric based on the Fourier spectrum
Autor(es) Gamazo Tejero, Ángel Javier
Abstract Recent progress in generative image modeling is leading to a new era of highresolution fakes visually indistinguishable from real life images. However, the development of metrics capable of discerning whether images are synthetic or not runs behind the race of achieving the best generator, thus bringing potential threats. We propose a rotation invariant metric capable of distinguishing real and generated images and prove its performance and correlation with subjective evaluation on a brain MRI dataset to generate synthetic white matter lesion images. We name this metric CSD (Circular Spectrum Distance) due to its circular nature and its inherent relation to the Fourier Spectrum. We find that this metric, as opposed to Frechet Inception Distance or Inception Score, detects overfitting during training in terms of generator memorisation without making use of any pretrained network. The conclusions are generalized to CelebA-HQ as a benchmark dataset.
Notas adicionales Trabajo de Fin de Máster. Máster Universitario en I.A. Avanzada: Fundamentos, Métodos y Aplicaciones. UNED
Materia(s) Ingeniería Informática
Palabra clave GAN metric
Fourier Spectrum
overfitting
memorisation
Editor(es) Universidad Nacional de Educación a Distancia (España). Escuela Técnica Superior de Ingeniería Informática. Departamento de Inteligencia Artificial
Director/Tutor Rincón Zamorano, Mariano
Cuadra Troncoso, José Manuel
Fecha 2020-09-28
Formato application/pdf
Identificador bibliuned:master-ETSInformatica-IAA-Ajgamazo
http://e-spacio.uned.es/fez/view/bibliuned:master-ETSInformatica-IAA-Ajgamazo
Idioma eng
Versión de la publicación acceptedVersion
Nivel de acceso y licencia http://creativecommons.org/licenses/by-nc-nd/4.0
info:eu-repo/semantics/openAccess
Tipo de recurso master Thesis
Tipo de acceso Acceso abierto

 
Versiones
Versión Tipo de filtro
Contador de citas: Google Scholar Search Google Scholar
Estadísticas de acceso: 297 Visitas, 206 Descargas  -  Estadísticas en detalle
Creado: Thu, 23 Sep 2021, 18:09:04 CET