Skew Class-Balanced Re-Weighting for Unbiased Scene Graph Generation

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 113
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorKang, Haeyongko
dc.contributor.authorYoo, Chang-Dongko
dc.date.accessioned2023-04-24T07:01:35Z-
dc.date.available2023-04-24T07:01:35Z-
dc.date.created2023-04-24-
dc.date.created2023-04-24-
dc.date.created2023-04-24-
dc.date.issued2023-03-
dc.identifier.citationMACHINE LEARNING AND KNOWLEDGE EXTRACTION, v.5, no.1, pp.287 - 303-
dc.identifier.issn2504-4990-
dc.identifier.urihttp://hdl.handle.net/10203/306378-
dc.description.abstractAn unbiased scene graph generation (SGG) algorithm referred to as Skew Class-Balanced Re-Weighting (SCR) is proposed for considering the unbiased predicate prediction caused by the long-tailed distribution. The prior works focus mainly on alleviating the deteriorating performances of the minority predicate predictions, showing drastic dropping recall scores, i.e., losing the majority predicate performances. It has not yet correctly analyzed the trade-off between majority and minority predicate performances in the limited SGG datasets. In this paper, to alleviate the issue, the Skew Class-Balanced Re-Weighting (SCR) loss function is considered for the unbiased SGG models. Leveraged by the skewness of biased predicate predictions, the SCR estimates the target predicate weight coefficient and then re-weights more to the biased predicates for better trading-off between the majority predicates and the minority ones. Extensive experiments conducted on the standard Visual Genome dataset and Open Image V4 and V6 show the performances and generality of the SCR with the traditional SGG models.-
dc.languageEnglish-
dc.publisherMDPI-
dc.titleSkew Class-Balanced Re-Weighting for Unbiased Scene Graph Generation-
dc.typeArticle-
dc.identifier.scopusid2-s2.0-85150983595-
dc.type.rimsART-
dc.citation.volume5-
dc.citation.issue1-
dc.citation.beginningpage287-
dc.citation.endingpage303-
dc.citation.publicationnameMACHINE LEARNING AND KNOWLEDGE EXTRACTION-
dc.identifier.doi10.3390/make5010018-
dc.contributor.localauthorYoo, Chang-Dong-
dc.description.isOpenAccessN-
dc.type.journalArticleArticle-
dc.subject.keywordAuthorscene graph generation (SGG)-
dc.subject.keywordAuthorskew class-balanced re-weighting (SCR)-
dc.subject.keywordAuthorpredicate sample estimates-
dc.subject.keywordAuthorskew class-balanced effective number-
dc.subject.keywordPlusCLASSIFICATION-
Appears in Collection
EE-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