schliessen

Filtern

 

Bibliotheken

DART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification

The accuracy of deep learning (e.g., convolutional neural networks) for an image classification task critically relies on the amount of labeled training data. Aiming to solve an image classification task on a new domain that lacks labeled data but gains access to cheaply available unlabeled data, un... Full description

Journal Title: arXiv.org Dec 30, 2018
Main Author: Fang, Xianghong
Other Authors: Bai, Haoli , Guo, Ziyi , Shen, Bin , Hoi, Steven , Xu, Zenglin
Format: Electronic Article Electronic Article
Language: English
Subjects:
Quelle: © ProQuest LLC All rights reserved
Zum Text:
SendSend as email Add to Book BagAdd to Book Bag
Staff View
recordid: proquest2162260520
title: DART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
format: Article
creator:
  • Fang, Xianghong
  • Bai, Haoli
  • Guo, Ziyi
  • Shen, Bin
  • Hoi, Steven
  • Xu, Zenglin
subjects:
  • Adaptation
  • State of the Art
  • Image Classification
  • Invariants
  • Adaptation
  • Machine Learning
  • Performance Evaluation
  • Domains
  • Artificial Neural Networks
  • Neural Networks
  • Image Contrast
  • Training
  • Classification
  • Computer Vision and Pattern Recognition
  • Machine Learning
ispartof: arXiv.org, Dec 30, 2018
description: The accuracy of deep learning (e.g., convolutional neural networks) for an image classification task critically relies on the amount of labeled training data. Aiming to solve an image classification task on a new domain that lacks labeled data but gains access to cheaply available unlabeled data, unsupervised domain adaptation is a promising technique to boost the performance without incurring extra labeling cost, by assuming images from different domains share some invariant characteristics. In this paper, we propose a new unsupervised domain adaptation method named Domain-Adversarial Residual-Transfer (DART) learning of Deep Neural Networks to tackle cross-domain image classification tasks. In contrast to the existing unsupervised domain adaption approaches, the proposed DART not only learns domain-invariant features via adversarial training, but also achieves robust domain-adaptive classification via a residual-transfer strategy, all in an end-to-end training framework. We evaluate the performance of the proposed method for cross-domain image classification tasks on several well-known benchmark data sets, in which our method clearly outperforms the state-of-the-art approaches.
language: eng
source: © ProQuest LLC All rights reserved
identifier:
fulltext: fulltext_linktorsrc
url: Link


@attributes
ID251630972
RANK0.07
NO1
SEARCH_ENGINEprimo_central_multiple_fe
SEARCH_ENGINE_TYPEPrimo Central Search Engine
LOCALfalse
PrimoNMBib
record
control
sourcerecordid2162260520
sourceidproquest
recordidTN_proquest2162260520
sourcesystemOther
pqid2162260520
display
typearticle
titleDART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
creatorFang, Xianghong ; Bai, Haoli ; Guo, Ziyi ; Shen, Bin ; Hoi, Steven ; Xu, Zenglin
contributorXu, Zenglin (pacrepositoryorg)
ispartofarXiv.org, Dec 30, 2018
subjectAdaptation ; State of the Art ; Image Classification ; Invariants ; Adaptation ; Machine Learning ; Performance Evaluation ; Domains ; Artificial Neural Networks ; Neural Networks ; Image Contrast ; Training ; Classification ; Computer Vision and Pattern Recognition ; Machine Learning
descriptionThe accuracy of deep learning (e.g., convolutional neural networks) for an image classification task critically relies on the amount of labeled training data. Aiming to solve an image classification task on a new domain that lacks labeled data but gains access to cheaply available unlabeled data, unsupervised domain adaptation is a promising technique to boost the performance without incurring extra labeling cost, by assuming images from different domains share some invariant characteristics. In this paper, we propose a new unsupervised domain adaptation method named Domain-Adversarial Residual-Transfer (DART) learning of Deep Neural Networks to tackle cross-domain image classification tasks. In contrast to the existing unsupervised domain adaption approaches, the proposed DART not only learns domain-invariant features via adversarial training, but also achieves robust domain-adaptive classification via a residual-transfer strategy, all in an end-to-end training framework. We evaluate the performance of the proposed method for cross-domain image classification tasks on several well-known benchmark data sets, in which our method clearly outperforms the state-of-the-art approaches.
languageeng
source
0© ProQuest LLC All rights reserved
1Engineering Database
2Publicly Available Content Database
3ProQuest Engineering Collection
4ProQuest Technology Collection
5ProQuest SciTech Collection
6Materials Science & Engineering Database
7ProQuest Central (new)
8ProQuest Central Korea
9SciTech Premium Collection
10Technology Collection
11ProQuest Central Essentials
oafree_for_read
links
openurl$$Topenurl_article
openurlfulltext$$Topenurlfull_article
linktorsrc$$Uhttp://search.proquest.com/docview/2162260520/?pq-origsite=primo$$EView_record_in_ProQuest_(subscribers_only)
search
creatorcontrib
0Fang, Xianghong
1Bai, Haoli
2Guo, Ziyi
3Shen, Bin
4Hoi, Steven
5Xu, Zenglin
titleDART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
descriptionThe accuracy of deep learning (e.g., convolutional neural networks) for an image classification task critically relies on the amount of labeled training data. Aiming to solve an image classification task on a new domain that lacks labeled data but gains access to cheaply available unlabeled data, unsupervised domain adaptation is a promising technique to boost the performance without incurring extra labeling cost, by assuming images from different domains share some invariant characteristics. In this paper, we propose a new unsupervised domain adaptation method named Domain-Adversarial Residual-Transfer (DART) learning of Deep Neural Networks to tackle cross-domain image classification tasks. In contrast to the existing unsupervised domain adaption approaches, the proposed DART not only learns domain-invariant features via adversarial training, but also achieves robust domain-adaptive classification via a residual-transfer strategy, all in an end-to-end training framework. We evaluate the performance of the proposed method for cross-domain image classification tasks on several well-known benchmark data sets, in which our method clearly outperforms the state-of-the-art approaches.
subject
0Adaptation
1State of the Art
2Image Classification
3Invariants
4Machine Learning
5Performance Evaluation
6Domains
7Artificial Neural Networks
8Neural Networks
9Image Contrast
10Training
11Classification
12Computer Vision and Pattern Recognition
general
0English
1Cornell University Library, arXiv.org
2Engineering Database
3Publicly Available Content Database
4ProQuest Engineering Collection
5ProQuest Technology Collection
6ProQuest SciTech Collection
7Materials Science & Engineering Database
8ProQuest Central (new)
9ProQuest Central Korea
10SciTech Premium Collection
11Technology Collection
12ProQuest Central Essentials
sourceidproquest
recordidproquest2162260520
rsrctypearticle
creationdate2018
addtitlearXiv.org
searchscope
01007853
11008875
21008886
31009127
41009240
510000041
610000053
710000120
810000250
910000255
1010000256
1110000258
1210000260
1310000265
1410000268
1510000281
1610000348
17proquest
scope
01007853
11008875
21008886
31009127
41009240
510000041
610000053
710000120
810000250
910000255
1010000256
1110000258
1210000260
1310000265
1410000268
1510000281
1610000348
17proquest
lsr43
01007853true
11008875true
21008886true
31009127true
41009240true
510000041true
610000053true
710000120true
810000250true
910000255true
1010000256true
1110000258true
1210000260true
1310000265true
1410000268true
1510000281true
1610000348true
startdate20181230
enddate20181230
sort
titleDART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
authorFang, Xianghong ; Bai, Haoli ; Guo, Ziyi ; Shen, Bin ; Hoi, Steven ; Xu, Zenglin
creationdate20181230
lso0120181230
facets
frbrgroupid7124767284292735159
frbrtype6
newrecords20190103
languageeng
creationdate2018
topic
0Adaptation
1State of the Art
2Image Classification
3Invariants
4Machine Learning
5Performance Evaluation
6Domains
7Artificial Neural Networks
8Neural Networks
9Image Contrast
10Training
11Classification
12Computer Vision and Pattern Recognition
collection
0Engineering Database
1Publicly Available Content Database
2ProQuest Engineering Collection
3ProQuest Technology Collection
4ProQuest SciTech Collection
5Materials Science & Engineering Database
6ProQuest Central (new)
7ProQuest Central Korea
8SciTech Premium Collection
9Technology Collection
10ProQuest Central Essentials
prefilterarticles
rsrctypearticles
creatorcontrib
0Fang, Xianghong
1Bai, Haoli
2Guo, Ziyi
3Shen, Bin
4Hoi, Steven
5Xu, Zenglin
jtitlearXiv.org
delivery
delcategoryRemote Search Resource
fulltextfulltext_linktorsrc
addata
aulast
0Fang
1Bai
2Guo
3Shen
4Hoi
5Xu
aufirst
0Xianghong
1Haoli
2Ziyi
3Bin
4Steven
5Zenglin
au
0Fang, Xianghong
1Bai, Haoli
2Guo, Ziyi
3Shen, Bin
4Hoi, Steven
5Xu, Zenglin
atitleDART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
jtitlearXiv.org
risdate20181230
formatjournal
genrearticle
ristypeJOUR
abstractThe accuracy of deep learning (e.g., convolutional neural networks) for an image classification task critically relies on the amount of labeled training data. Aiming to solve an image classification task on a new domain that lacks labeled data but gains access to cheaply available unlabeled data, unsupervised domain adaptation is a promising technique to boost the performance without incurring extra labeling cost, by assuming images from different domains share some invariant characteristics. In this paper, we propose a new unsupervised domain adaptation method named Domain-Adversarial Residual-Transfer (DART) learning of Deep Neural Networks to tackle cross-domain image classification tasks. In contrast to the existing unsupervised domain adaption approaches, the proposed DART not only learns domain-invariant features via adversarial training, but also achieves robust domain-adaptive classification via a residual-transfer strategy, all in an end-to-end training framework. We evaluate the performance of the proposed method for cross-domain image classification tasks on several well-known benchmark data sets, in which our method clearly outperforms the state-of-the-art approaches.
copIthaca
pubCornell University Library, arXiv.org
urlhttp://search.proquest.com/docview/2162260520/
oafree_for_read
date2018-12-30