A Simple Framework for Robust Out-of-Distribution Detection

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 48
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorHur, Youngbumko
dc.contributor.authorYang, Eunhoko
dc.contributor.authorHwang, Sung Juko
dc.date.accessioned2022-04-15T06:51:06Z-
dc.date.available2022-04-15T06:51:06Z-
dc.date.created2022-03-21-
dc.date.issued2022-
dc.identifier.citationIEEE ACCESS, v.10, pp.23086 - 23097-
dc.identifier.issn2169-3536-
dc.identifier.urihttp://hdl.handle.net/10203/294827-
dc.description.abstractOut-of-distribution (OOD) detection, i.e., identifying whether a given test sample is drawn from outside the training distribution, is essential for a deep classifier to be deployed in a real-world application. The existing state-of-the-art methods of OOD detection tackle this issue by utilizing the internal feature of the classification network. However, we found that such detection methods inherently struggle to detect hard OOD images, i.e., drawn near from the training distribution: a naive softmax-based baseline even outperforms them. Motivated by this, we propose a simple yet effective training scheme for further calibrating the softmax probability of a classifier to achieve high OOD detection performance under both hard and easy scenarios. In particular, we suggest to optimize consistency regularization and self-supervised loss during training. Our experiments demonstrate the superiority of our simple method under various OOD detection scenarios.-
dc.languageEnglish-
dc.publisherIEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC-
dc.titleA Simple Framework for Robust Out-of-Distribution Detection-
dc.typeArticle-
dc.identifier.wosid000764617900001-
dc.identifier.scopusid2-s2.0-85125347599-
dc.type.rimsART-
dc.citation.volume10-
dc.citation.beginningpage23086-
dc.citation.endingpage23097-
dc.citation.publicationnameIEEE ACCESS-
dc.identifier.doi10.1109/ACCESS.2022.3153723-
dc.contributor.localauthorYang, Eunho-
dc.contributor.localauthorHwang, Sung Ju-
dc.contributor.nonIdAuthorHur, Youngbum-
dc.description.isOpenAccessN-
dc.type.journalArticleArticle-
dc.subject.keywordAuthorTraining-
dc.subject.keywordAuthorCalibration-
dc.subject.keywordAuthorTask analysis-
dc.subject.keywordAuthorFeature extraction-
dc.subject.keywordAuthorStandards-
dc.subject.keywordAuthorNeural networks-
dc.subject.keywordAuthorLicenses-
dc.subject.keywordAuthorOut-of-distribution detection-
dc.subject.keywordAuthornetwork calibration-
dc.subject.keywordAuthordeep neural networks-
dc.subject.keywordAuthorconsistency regularization-
dc.subject.keywordAuthorself-supervised learning-
Appears in Collection
AI-Journal Papers(저널논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0