DC Field | Value | Language |
---|---|---|
dc.contributor.author | Nguyen, Thanh | ko |
dc.contributor.author | Pham, Trung Xuan | ko |
dc.contributor.author | Zhang, Chaoning | ko |
dc.contributor.author | Luu, Tung M. | ko |
dc.contributor.author | Vu, Thang | ko |
dc.contributor.author | Yoo, Chang-Dong | ko |
dc.date.accessioned | 2023-04-03T05:02:06Z | - |
dc.date.available | 2023-04-03T05:02:06Z | - |
dc.date.created | 2023-04-03 | - |
dc.date.created | 2023-04-03 | - |
dc.date.created | 2023-04-03 | - |
dc.date.issued | 2023 | - |
dc.identifier.citation | IEEE ACCESS, v.11, pp.21534 - 21545 | - |
dc.identifier.issn | 2169-3536 | - |
dc.identifier.uri | http://hdl.handle.net/10203/305959 | - |
dc.description.abstract | Self-supervised learning (SSL) has gained remarkable success, for which contrastive learning (CL) plays a key role. However, the recent development of new non-CL frameworks has achieved comparable or better performance with high improvement potential, prompting researchers to enhance these frameworks further. Assimilating CL into non-CL frameworks has been thought to be beneficial, but empirical evidence indicates no visible improvements. In view of that, this paper proposes a strategy of performing CL along the dimensional direction instead of along the batch direction as done in conventional contrastive learning, named Dimensional Contrastive Learning (DimCL). DimCL aims to enhance the feature diversity, and it can serve as a regularizer to prior SSL frameworks. DimCL has been found to be effective, and the hardness-aware property is identified as a critical reason for its success. Extensive experimental results reveal that assimilating DimCL into SSL frameworks leads to performance improvement by a non-trivial margin on various datasets and backbone architectures. | - |
dc.language | English | - |
dc.publisher | IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC | - |
dc.title | DimCL: Dimensional Contrastive Learning for Improving Self-Supervised Learning | - |
dc.type | Article | - |
dc.identifier.wosid | 000946224900001 | - |
dc.identifier.scopusid | 2-s2.0-85150414171 | - |
dc.type.rims | ART | - |
dc.citation.volume | 11 | - |
dc.citation.beginningpage | 21534 | - |
dc.citation.endingpage | 21545 | - |
dc.citation.publicationname | IEEE ACCESS | - |
dc.identifier.doi | 10.1109/ACCESS.2023.3236087 | - |
dc.contributor.localauthor | Yoo, Chang-Dong | - |
dc.contributor.nonIdAuthor | Nguyen, Thanh | - |
dc.contributor.nonIdAuthor | Pham, Trung Xuan | - |
dc.contributor.nonIdAuthor | Luu, Tung M. | - |
dc.description.isOpenAccess | N | - |
dc.type.journalArticle | Article | - |
dc.subject.keywordAuthor | Transfer learning | - |
dc.subject.keywordAuthor | Task analysis | - |
dc.subject.keywordAuthor | Self-supervised learning | - |
dc.subject.keywordAuthor | Learning systems | - |
dc.subject.keywordAuthor | Loss measurement | - |
dc.subject.keywordAuthor | Computer vision | - |
dc.subject.keywordAuthor | Self-supervise learning | - |
dc.subject.keywordAuthor | computer vision | - |
dc.subject.keywordAuthor | contrastive learning | - |
dc.subject.keywordAuthor | deep learning | - |
dc.subject.keywordAuthor | transfer learning | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.