Explicit Use of Fourier Spectrum in Generative Adversarial Networks

Simple item page
dc.contributor.advisorJiang, Hui
dc.contributor.authorSheikh Gargar, Soroush
dc.date.accessioned2022-03-03T13:58:07Z
dc.date.available2022-03-03T13:58:07Z
dc.date.copyright2021-09
dc.date.issued2022-03-03
dc.date.updated2022-03-03T13:58:07Z
dc.degree.disciplineElectrical and Computer Engineering
dc.degree.levelMaster's
dc.degree.nameMASc - Master of Applied Science
dc.description.abstractGenerative Adversarial Networks have got the researcher's attention due to their state-of- the-art performance in generating new images with only a dataset of the target distribution. It has been shown that there is a dissimilarity between the spectrum of authentic images and fake ones. Since the Fourier transform is a bijective mapping, saying that the model has a significant problem in learning the original distribution is a fair conclusion. In this work, we investigate the possible reasons for the mentioned drawback in the architecture and mathematical theory of the current GANs. Then we propose a new model reducing the discrepancies between the spectrum of the actual and fake images. To that end, we design a brand new architecture for the frequency domain using the blueprint of geometric deep learning. Then, we experimentally show that promising improvements in the quality of the generated images by considering the Fourier domain representation of the original data as a principal feature in the training process.
dc.identifier.urihttp://hdl.handle.net/10315/39062
dc.languageen
dc.rightsAuthor owns copyright, except where explicitly noted. Please contact the author directly with licensing requests.
dc.subjectComputer engineering
dc.subject.keywordsGenerative adversarial networks
dc.subject.keywordsDeep learning
dc.subject.keywordsMachine learning
dc.subject.keywordsGeometric deep learning
dc.subject.keywordsNeural networks
dc.subject.keywordsComputer vision
dc.subject.keywordsGenerative models
dc.subject.keywordsFrequency domain
dc.subject.keywordsFrequency
dc.subject.keywordsFourier transform
dc.subject.keywordsFourier
dc.subject.keywordsConvolutional neural networks
dc.subject.keywordsCNN
dc.subject.keywordsGAN
dc.subject.keywordsCheckerboard patterns
dc.subject.keywordsFrequency correction
dc.subject.keywordsFrequency driven generative models
dc.titleExplicit Use of Fourier Spectrum in Generative Adversarial Networks
dc.typeElectronic Thesis or Dissertation

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
SheikhGargar_Soroush_2021_Masters.pdf
Size:
5.1 MB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 2 of 2
No Thumbnail Available
Name:
license.txt
Size:
1.87 KB
Format:
Plain Text
Description:
Download
No Thumbnail Available
Name:
YorkU_ETDlicense.txt
Size:
3.39 KB
Format:
Plain Text
Description:
Download

Collections

Electrical and Computer Engineering

All items in the YorkSpace institutional repository are protected by copyright, with all rights reserved except where explicitly noted.