Versatile Auxiliary Classification and Regression With Generative Adversarial Networks
One of the most interesting challenges in Artificial Intelligence is to train conditional generators which are able to provide labeled adversarial samples drawn from a specific distribution. For a successful implementation of conditional generators, the created samples are constrained to a specific...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2021-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9369348/ |
id |
doaj-c7c57e7519be44a0a116e22ec9392508 |
---|---|
record_format |
Article |
spelling |
doaj-c7c57e7519be44a0a116e22ec93925082021-03-30T15:21:21ZengIEEEIEEE Access2169-35362021-01-019388103882510.1109/ACCESS.2021.30637939369348Versatile Auxiliary Classification and Regression With Generative Adversarial NetworksShabab Bazrafkan0Viktor Varkarakis1https://orcid.org/0000-0002-3877-2802Joseph Lemley2https://orcid.org/0000-0002-0595-2313Hossein Javidnia3Peter Corcoran4https://orcid.org/0000-0003-1670-4793Department of Physics, University of Antwerp, Antwerp, BelgiumCollege of Engineering and Informatics, National University of Ireland Galway, Galway, IrelandCollege of Engineering and Informatics, National University of Ireland Galway, Galway, IrelandSchool of Computer Science and Statistics ADAPT Centre, Trinity College Dublin, Dublin 2, IrelandCollege of Engineering and Informatics, National University of Ireland Galway, Galway, IrelandOne of the most interesting challenges in Artificial Intelligence is to train conditional generators which are able to provide labeled adversarial samples drawn from a specific distribution. For a successful implementation of conditional generators, the created samples are constrained to a specific class. In this work, a new framework is presented to train a deep conditional generator by placing a classifier or regression model in parallel with the discriminator and back propagate the classification or regression error through the generator network. Special cases for binary classification, multi-class classification, and regression are studied. Experimental results on several data-sets are provided and the results are compared with similar state-of-the-art techniques. The main advantage of the method is that it is versatile and applicable to any variation of Generative Adversarial Network (GAN) implementation but also it is shown to obtain superior results compared to other methods. The mathematical proofs for the proposed scheme for both classification and regression are presented.https://ieeexplore.ieee.org/document/9369348/Conditional generatorsdeep neural networksgenerative adversarial networks |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Shabab Bazrafkan Viktor Varkarakis Joseph Lemley Hossein Javidnia Peter Corcoran |
spellingShingle |
Shabab Bazrafkan Viktor Varkarakis Joseph Lemley Hossein Javidnia Peter Corcoran Versatile Auxiliary Classification and Regression With Generative Adversarial Networks IEEE Access Conditional generators deep neural networks generative adversarial networks |
author_facet |
Shabab Bazrafkan Viktor Varkarakis Joseph Lemley Hossein Javidnia Peter Corcoran |
author_sort |
Shabab Bazrafkan |
title |
Versatile Auxiliary Classification and Regression With Generative Adversarial Networks |
title_short |
Versatile Auxiliary Classification and Regression With Generative Adversarial Networks |
title_full |
Versatile Auxiliary Classification and Regression With Generative Adversarial Networks |
title_fullStr |
Versatile Auxiliary Classification and Regression With Generative Adversarial Networks |
title_full_unstemmed |
Versatile Auxiliary Classification and Regression With Generative Adversarial Networks |
title_sort |
versatile auxiliary classification and regression with generative adversarial networks |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2021-01-01 |
description |
One of the most interesting challenges in Artificial Intelligence is to train conditional generators which are able to provide labeled adversarial samples drawn from a specific distribution. For a successful implementation of conditional generators, the created samples are constrained to a specific class. In this work, a new framework is presented to train a deep conditional generator by placing a classifier or regression model in parallel with the discriminator and back propagate the classification or regression error through the generator network. Special cases for binary classification, multi-class classification, and regression are studied. Experimental results on several data-sets are provided and the results are compared with similar state-of-the-art techniques. The main advantage of the method is that it is versatile and applicable to any variation of Generative Adversarial Network (GAN) implementation but also it is shown to obtain superior results compared to other methods. The mathematical proofs for the proposed scheme for both classification and regression are presented. |
topic |
Conditional generators deep neural networks generative adversarial networks |
url |
https://ieeexplore.ieee.org/document/9369348/ |
work_keys_str_mv |
AT shababbazrafkan versatileauxiliaryclassificationandregressionwithgenerativeadversarialnetworks AT viktorvarkarakis versatileauxiliaryclassificationandregressionwithgenerativeadversarialnetworks AT josephlemley versatileauxiliaryclassificationandregressionwithgenerativeadversarialnetworks AT hosseinjavidnia versatileauxiliaryclassificationandregressionwithgenerativeadversarialnetworks AT petercorcoran versatileauxiliaryclassificationandregressionwithgenerativeadversarialnetworks |
_version_ |
1724179666850283520 |