Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks
| Title: | Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks |
|---|---|
| Authors: | Ding, Sihao, Wallin, Andreas |
| Source: | Pattern Recognition Letters, vol. 122, pp. 66-72, 1 May 2019 |
| Publication Year: | 2017 |
| Collection: | Computer Science |
| Subject Terms: | Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning |
| More Details: | A conditional Generative Adversarial Network allows for generating samples conditioned on certain external information. Being able to recover latent and conditional vectors from a condi- tional GAN can be potentially valuable in various applications, ranging from image manipulation for entertaining purposes to diagnosis of the neural networks for security purposes. In this work, we show that it is possible to recover both latent and conditional vectors from generated images given the generator of a conditional generative adversarial network. Such a recovery is not trivial due to the often multi-layered non-linearity of deep neural networks. Furthermore, the effect of such recovery applied on real natural images are investigated. We discovered that there exists a gap between the recovery performance on generated and real images, which we believe comes from the difference between generated data distribution and real data distribution. Experiments are conducted to evaluate the recovered conditional vectors and the reconstructed images from these recovered vectors quantitatively and qualitatively, showing promising results. Comment: Under consideration for Pattern Recognition Letters, 11 pages |
| Document Type: | Working Paper |
| DOI: | 10.1016/j.patrec.2019.02.020 |
| Access URL: | http://arxiv.org/abs/1712.01833 |
| Accession Number: | edsarx.1712.01833 |
| Database: | arXiv |
| FullText | Text: Availability: 0 CustomLinks: – Url: http://arxiv.org/abs/1712.01833 Name: EDS - Arxiv Category: fullText Text: View this record from Arxiv MouseOverText: View this record from Arxiv – Url: https://resolver.ebsco.com/c/xy5jbn/result?sid=EBSCO:edsarx&genre=article&issn=&ISBN=&volume=&issue=&date=20171205&spage=&pages=&title=Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks&atitle=Towards%20Recovery%20of%20Conditional%20Vectors%20from%20Conditional%20Generative%20Adversarial%20Networks&aulast=Ding%2C%20Sihao&id=DOI:10.1016/j.patrec.2019.02.020 Name: Full Text Finder (for New FTF UI) (s8985755) Category: fullText Text: Find It @ SCU Libraries MouseOverText: Find It @ SCU Libraries |
|---|---|
| Header | DbId: edsarx DbLabel: arXiv An: edsarx.1712.01833 RelevancyScore: 971 AccessLevel: 3 PubType: Report PubTypeId: report PreciseRelevancyScore: 970.702392578125 |
| IllustrationInfo | |
| Items | – Name: Title Label: Title Group: Ti Data: Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks – Name: Author Label: Authors Group: Au Data: <searchLink fieldCode="AR" term="%22Ding%2C+Sihao%22">Ding, Sihao</searchLink><br /><searchLink fieldCode="AR" term="%22Wallin%2C+Andreas%22">Wallin, Andreas</searchLink> – Name: TitleSource Label: Source Group: Src Data: Pattern Recognition Letters, vol. 122, pp. 66-72, 1 May 2019 – Name: DatePubCY Label: Publication Year Group: Date Data: 2017 – Name: Subset Label: Collection Group: HoldingsInfo Data: Computer Science – Name: Subject Label: Subject Terms Group: Su Data: <searchLink fieldCode="DE" term="%22Computer+Science+-+Computer+Vision+and+Pattern+Recognition%22">Computer Science - Computer Vision and Pattern Recognition</searchLink><br /><searchLink fieldCode="DE" term="%22Computer+Science+-+Machine+Learning%22">Computer Science - Machine Learning</searchLink> – Name: Abstract Label: Description Group: Ab Data: A conditional Generative Adversarial Network allows for generating samples conditioned on certain external information. Being able to recover latent and conditional vectors from a condi- tional GAN can be potentially valuable in various applications, ranging from image manipulation for entertaining purposes to diagnosis of the neural networks for security purposes. In this work, we show that it is possible to recover both latent and conditional vectors from generated images given the generator of a conditional generative adversarial network. Such a recovery is not trivial due to the often multi-layered non-linearity of deep neural networks. Furthermore, the effect of such recovery applied on real natural images are investigated. We discovered that there exists a gap between the recovery performance on generated and real images, which we believe comes from the difference between generated data distribution and real data distribution. Experiments are conducted to evaluate the recovered conditional vectors and the reconstructed images from these recovered vectors quantitatively and qualitatively, showing promising results.<br />Comment: Under consideration for Pattern Recognition Letters, 11 pages – Name: TypeDocument Label: Document Type Group: TypDoc Data: Working Paper – Name: DOI Label: DOI Group: ID Data: 10.1016/j.patrec.2019.02.020 – Name: URL Label: Access URL Group: URL Data: <link linkTarget="URL" linkTerm="http://arxiv.org/abs/1712.01833" linkWindow="_blank">http://arxiv.org/abs/1712.01833</link> – Name: AN Label: Accession Number Group: ID Data: edsarx.1712.01833 |
| PLink | https://login.libproxy.scu.edu/login?url=https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&scope=site&db=edsarx&AN=edsarx.1712.01833 |
| RecordInfo | BibRecord: BibEntity: Identifiers: – Type: doi Value: 10.1016/j.patrec.2019.02.020 Subjects: – SubjectFull: Computer Science - Computer Vision and Pattern Recognition Type: general – SubjectFull: Computer Science - Machine Learning Type: general Titles: – TitleFull: Towards Recovery of Conditional Vectors from Conditional Generative Adversarial Networks Type: main BibRelationships: HasContributorRelationships: – PersonEntity: Name: NameFull: Ding, Sihao – PersonEntity: Name: NameFull: Wallin, Andreas IsPartOfRelationships: – BibEntity: Dates: – D: 05 M: 12 Type: published Y: 2017 |
| ResultId | 1 |