Virtual adversarial training in feature space to improve unsupervised video domain adaptation. Gorpincenko, A., French, G., & Mackiewicz, M. Electronic Imaging, 2021(10):258–1-258-6, January, 2021. ![Virtual adversarial training in feature space to improve unsupervised video domain adaptation [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Virtual Adversarial Training has recently seen a lot of success in semi-supervised learning, as well as unsupervised Domain Adaptation. However, so far it has been used on input samples in the pixel space, whereas we propose to apply it directly to feature vectors. We also discuss the unstable behaviour of entropy minimization and Decision-Boundary Iterative Refinement Training With a Teacher in Domain Adaptation, and suggest substitutes that achieve similar behaviour. By adding the aforementioned techniques to the state of the art model TA3N, we either maintain competitive results or outperform prior art in multiple unsupervised video Domain Adaptation tasks.
@article{uea81506,
volume = {2021},
number = {10},
journal = {Electronic Imaging},
month = {January},
title = {Virtual adversarial training in feature space to improve unsupervised video domain adaptation},
doi = {10.2352/ISSN.2470-1173.2021.10.IPAS-258},
year = {2021},
pages = {258--1-258-6},
issn = {2470-1173},
abstract = {Virtual Adversarial Training has recently seen a lot of success in semi-supervised learning, as well as unsupervised Domain Adaptation. However, so far it has been used on input samples in the pixel space, whereas we propose to apply it directly to feature vectors. We also discuss the unstable behaviour of entropy minimization and Decision-Boundary Iterative Refinement Training With a Teacher in Domain Adaptation, and suggest substitutes that achieve similar behaviour. By adding the aforementioned techniques to the state of the art model TA3N, we either maintain competitive results or outperform prior art in multiple unsupervised video Domain Adaptation tasks.},
author = {Gorpincenko, Artjoms and French, Geoffrey and Mackiewicz, Michal},
keywords = {computer graphics and computer-aided design,computer science applications,human-computer interaction,software,electrical and electronic engineering,atomic and molecular physics, and optics ,/dk/atira/pure/subjectarea/asjc/1700/1704},
url = {https://ueaeprints.uea.ac.uk/id/eprint/81506/}
}
Downloads: 0
{"_id":"mnTJipw6QBpm4WQ8L","bibbaseid":"gorpincenko-french-mackiewicz-virtualadversarialtraininginfeaturespacetoimproveunsupervisedvideodomainadaptation-2021","author_short":["Gorpincenko, A.","French, G.","Mackiewicz, M."],"bibdata":{"bibtype":"article","type":"article","volume":"2021","number":"10","journal":"Electronic Imaging","month":"January","title":"Virtual adversarial training in feature space to improve unsupervised video domain adaptation","doi":"10.2352/ISSN.2470-1173.2021.10.IPAS-258","year":"2021","pages":"258–1-258-6","issn":"2470-1173","abstract":"Virtual Adversarial Training has recently seen a lot of success in semi-supervised learning, as well as unsupervised Domain Adaptation. However, so far it has been used on input samples in the pixel space, whereas we propose to apply it directly to feature vectors. We also discuss the unstable behaviour of entropy minimization and Decision-Boundary Iterative Refinement Training With a Teacher in Domain Adaptation, and suggest substitutes that achieve similar behaviour. By adding the aforementioned techniques to the state of the art model TA3N, we either maintain competitive results or outperform prior art in multiple unsupervised video Domain Adaptation tasks.","author":[{"propositions":[],"lastnames":["Gorpincenko"],"firstnames":["Artjoms"],"suffixes":[]},{"propositions":[],"lastnames":["French"],"firstnames":["Geoffrey"],"suffixes":[]},{"propositions":[],"lastnames":["Mackiewicz"],"firstnames":["Michal"],"suffixes":[]}],"keywords":"computer graphics and computer-aided design,computer science applications,human-computer interaction,software,electrical and electronic engineering,atomic and molecular physics, and optics ,/dk/atira/pure/subjectarea/asjc/1700/1704","url":"https://ueaeprints.uea.ac.uk/id/eprint/81506/","bibtex":"@article{uea81506,\n volume = {2021},\n number = {10},\n journal = {Electronic Imaging},\n month = {January},\n title = {Virtual adversarial training in feature space to improve unsupervised video domain adaptation},\n doi = {10.2352/ISSN.2470-1173.2021.10.IPAS-258},\n year = {2021},\n pages = {258--1-258-6},\n issn = {2470-1173},\n abstract = {Virtual Adversarial Training has recently seen a lot of success in semi-supervised learning, as well as unsupervised Domain Adaptation. However, so far it has been used on input samples in the pixel space, whereas we propose to apply it directly to feature vectors. We also discuss the unstable behaviour of entropy minimization and Decision-Boundary Iterative Refinement Training With a Teacher in Domain Adaptation, and suggest substitutes that achieve similar behaviour. By adding the aforementioned techniques to the state of the art model TA3N, we either maintain competitive results or outperform prior art in multiple unsupervised video Domain Adaptation tasks.},\n author = {Gorpincenko, Artjoms and French, Geoffrey and Mackiewicz, Michal},\n keywords = {computer graphics and computer-aided design,computer science applications,human-computer interaction,software,electrical and electronic engineering,atomic and molecular physics, and optics ,/dk/atira/pure/subjectarea/asjc/1700/1704},\n url = {https://ueaeprints.uea.ac.uk/id/eprint/81506/}\n}\n\n","author_short":["Gorpincenko, A.","French, G.","Mackiewicz, M."],"key":"uea81506","id":"uea81506","bibbaseid":"gorpincenko-french-mackiewicz-virtualadversarialtraininginfeaturespacetoimproveunsupervisedvideodomainadaptation-2021","role":"author","urls":{"Paper":"https://ueaeprints.uea.ac.uk/id/eprint/81506/"},"keyword":["computer graphics and computer-aided design","computer science applications","human-computer interaction","software","electrical and electronic engineering","atomic and molecular physics","and optics","/dk/atira/pure/subjectarea/asjc/1700/1704"],"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://ueaeprints.uea.ac.uk/cgi/search/archive/advanced/export_uea_BibTeX.bib?dataset=archive&screen=Search&_action_export=1&output=BibTeX&exp=0%7C1%7C-date%2Fcreators_name%2Ftitle%7Carchive%7C-%7Ccreators_search_name%3Acreators_search_name%3AANY%3AEQ%3A+Wang%2C+Wenjia+Harvey%2C+R+Finlayson%2C+Graham+Mackiewicz%2C+Michal++De+la+Iglesia%2C+B%7Cdate%3Adate%3AALL%3AEQ%3A2021-%7Cdivisions%3Adivisions%3AANY%3AEQ%3ACMP%7C-%7Ceprint_status%3Aeprint_status%3AANY%3AEQ%3Aarchive%7Cmetadata_visibility%3Ametadata_visibility%3AANY%3AEQ%3Ashow&n=&cache=8778568","dataSources":["sad4FokpAYKGF23qb"],"keywords":["computer graphics and computer-aided design","computer science applications","human-computer interaction","software","electrical and electronic engineering","atomic and molecular physics","and optics","/dk/atira/pure/subjectarea/asjc/1700/1704"],"search_terms":["virtual","adversarial","training","feature","space","improve","unsupervised","video","domain","adaptation","gorpincenko","french","mackiewicz"],"title":"Virtual adversarial training in feature space to improve unsupervised video domain adaptation","year":2021}