File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
  • Find via Find It@HKUL
Supplementary

Conference Paper: Differentiable dynamic normalization for learning deep representation

TitleDifferentiable dynamic normalization for learning deep representation
Authors
Luo, PPang, ZShao, WZhang, RRen, JWu, L
KeywordsAlgorithms
Computer Vision
Issue Date2019
PublisherPMLR. The Journal's web site is located at http://proceedings.mlr.press/
Citation
The 36th International Conference on Machine Learning (ICML 2019), Long Beach, CA, USA, 10-15 June 2019. In Proceedings of Machine Learning Research (PMLR), v. 97, p. 4203-4211 How to Cite?
AbstractThis work presents Dynamic Normalization (DN), which is able to learn arbitrary normalization operations for different convolutional layers in a deep ConvNet. Unlike existing normalization approaches that predefined computations of the statistics (mean and variance), DN learns to estimate them. DN has several appealing benefits. First, it adapts to various networks, tasks, and batch sizes. Second, it can be easily implemented and trained in a differentiable end-to-end manner with merely small number of parameters. Third, its matrix formulation represents a wide range of normalization methods, shedding light on analyzing them theoretically. Extensive studies show that DN outperforms its counterparts in CIFAR10 and ImageNet.
DescriptionPoster Presentation
Persistent Identifierhttp://hdl.handle.net/10722/284152
ISSN
2640-3498

 

DC FieldValueLanguage
dc.contributor.authorLuo, P-
dc.contributor.authorPang, Z-
dc.contributor.authorShao, W-
dc.contributor.authorZhang, R-
dc.contributor.authorRen, J-
dc.contributor.authorWu, L-
dc.date.accessioned2020-07-20T05:56:30Z-
dc.date.available2020-07-20T05:56:30Z-
dc.date.issued2019-
dc.identifier.citationThe 36th International Conference on Machine Learning (ICML 2019), Long Beach, CA, USA, 10-15 June 2019. In Proceedings of Machine Learning Research (PMLR), v. 97, p. 4203-4211-
dc.identifier.issn2640-3498-
dc.identifier.urihttp://hdl.handle.net/10722/284152-
dc.descriptionPoster Presentation-
dc.description.abstractThis work presents Dynamic Normalization (DN), which is able to learn arbitrary normalization operations for different convolutional layers in a deep ConvNet. Unlike existing normalization approaches that predefined computations of the statistics (mean and variance), DN learns to estimate them. DN has several appealing benefits. First, it adapts to various networks, tasks, and batch sizes. Second, it can be easily implemented and trained in a differentiable end-to-end manner with merely small number of parameters. Third, its matrix formulation represents a wide range of normalization methods, shedding light on analyzing them theoretically. Extensive studies show that DN outperforms its counterparts in CIFAR10 and ImageNet.-
dc.languageeng-
dc.publisherPMLR. The Journal's web site is located at http://proceedings.mlr.press/-
dc.relation.ispartofThe 36th International Conference on Machine Learning (ICML)-
dc.relation.ispartofProceedings of Machine Learning Research (PMLR)-
dc.subjectAlgorithms-
dc.subjectComputer Vision-
dc.titleDifferentiable dynamic normalization for learning deep representation-
dc.typeConference_Paper-
dc.identifier.emailLuo, P: pluo@hku.hk-
dc.identifier.authorityLuo, P=rp02575-
dc.identifier.hkuros311011-
dc.identifier.volume97-
dc.identifier.spage4203-
dc.identifier.epage4211-
dc.publisher.placeUnited States-
dc.identifier.issnl2640-3498-

Export via OAI-PMH Interface in XML Formats

OR

Export to Other Non-XML Formats