DC Field | Value | Language |
---|---|---|
dc.contributor.author | Yoo, Byungin | ko |
dc.contributor.author | Sylvain, Tristan | ko |
dc.contributor.author | Bengio, Yoshua | ko |
dc.contributor.author | Kim, Junmo | ko |
dc.date.accessioned | 2021-01-04T08:50:12Z | - |
dc.date.available | 2021-01-04T08:50:12Z | - |
dc.date.created | 2021-01-04 | - |
dc.date.created | 2021-01-04 | - |
dc.date.created | 2021-01-04 | - |
dc.date.issued | 2020 | - |
dc.identifier.citation | IEEE ACCESS, v.8, pp.219160 - 219173 | - |
dc.identifier.issn | 2169-3536 | - |
dc.identifier.uri | http://hdl.handle.net/10203/279458 | - |
dc.description.abstract | In this paper, we propose a Generative Translation Classification Network (GTCN) for improving visual classification accuracy in settings where classes are visually similar and data is scarce. For this purpose, we propose joint learning from a scratch to train a classifier and a generative stochastic translation network end-to-end. The translation network is used to perform on-line data augmentation across classes, whereas previous works have mostly involved domain adaptation. To help the model further benefit from this data-augmentation, we introduce an adaptive fade-in loss and a quadruplet loss. We perform experiments on multiple datasets to demonstrate the proposed method's performance in varied settings. Of particular interest, training on 40% of the dataset is enough for our model to surpass the performance of baselines trained on the full dataset. When our architecture is trained on the full dataset, we achieve comparable performance with state-of-the-art methods despite using a light-weight architecture. | - |
dc.language | English | - |
dc.publisher | IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC | - |
dc.title | Joint Learning of Generative Translator and Classifier for Visually Similar Classes | - |
dc.type | Article | - |
dc.identifier.wosid | 000597792200001 | - |
dc.identifier.scopusid | 2-s2.0-85097774811 | - |
dc.type.rims | ART | - |
dc.citation.volume | 8 | - |
dc.citation.beginningpage | 219160 | - |
dc.citation.endingpage | 219173 | - |
dc.citation.publicationname | IEEE ACCESS | - |
dc.identifier.doi | 10.1109/ACCESS.2020.3042302 | - |
dc.contributor.localauthor | Kim, Junmo | - |
dc.contributor.nonIdAuthor | Sylvain, Tristan | - |
dc.contributor.nonIdAuthor | Bengio, Yoshua | - |
dc.description.isOpenAccess | Y | - |
dc.type.journalArticle | Article | - |
dc.subject.keywordAuthor | Training | - |
dc.subject.keywordAuthor | Data models | - |
dc.subject.keywordAuthor | Training data | - |
dc.subject.keywordAuthor | Visualization | - |
dc.subject.keywordAuthor | Adaptation models | - |
dc.subject.keywordAuthor | Semisupervised learning | - |
dc.subject.keywordAuthor | Faces | - |
dc.subject.keywordAuthor | Artificial neural networks | - |
dc.subject.keywordAuthor | feature extraction | - |
dc.subject.keywordAuthor | image classification | - |
dc.subject.keywordAuthor | image generation | - |
dc.subject.keywordAuthor | pattern analysis | - |
dc.subject.keywordAuthor | semisupervised learning | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.