View in EDS HTML Full Text PDF Full Text

Variational Rectification Inference for Learning with Noisy Labels.

Saved in:
Bibliographic Details
Title: Variational Rectification Inference for Learning with Noisy Labels.
Authors: Sun, Haoliang1 (AUTHOR) haolsun@sdu.edu.cn, Wei, Qi2 (AUTHOR) 1998v7@gmail.com, Feng, Lei3 (AUTHOR) feng_lei@sutd.edu.sg, Hu, Yupeng1 (AUTHOR) huyupeng@sdu.edu.cn, Liu, Fan4 (AUTHOR) liufancs@gmail.com, Fan, Hehe5 (AUTHOR) hehefan@zju.edu.cn, Yin, Yilong1 (AUTHOR) ylyin@sdu.edu.cn
Source: International Journal of Computer Vision. Feb2025, Vol. 133 Issue 2, p652-671. 20p.
Subjects: Dirac function, Artificial intelligence, Machine learning, Bilevel programming, Latent variables
Abstract: Label noise has been broadly observed in real-world datasets. To mitigate the negative impact of overfitting to label noise for deep models, effective strategies (e.g., re-weighting, or loss rectification) have been broadly applied in prevailing approaches, which have been generally learned under the meta-learning scenario. Despite the robustness of noise achieved by the probabilistic meta-learning models, they usually suffer from model collapse that degenerates generalization performance. In this paper, we propose variational rectification inference (VRI) to formulate the adaptive rectification for loss functions as an amortized variational inference problem and derive the evidence lower bound under the meta-learning framework. Specifically, VRI is constructed as a hierarchical Bayes by treating the rectifying vector as a latent variable, which can rectify the loss of the noisy sample with the extra randomness regularization and is, therefore, more robust to label noise. To achieve the inference of the rectifying vector, we approximate its conditional posterior with an amortization meta-network. By introducing the variational term in VRI, the conditional posterior is estimated accurately and avoids collapsing to a Dirac delta function, which can significantly improve the generalization performance. The elaborated meta-network and prior network adhere to the smoothness assumption, enabling the generation of reliable rectification vectors. Given a set of clean meta-data, VRI can be efficiently meta-learned within the bi-level optimization programming. Besides, theoretical analysis guarantees that the meta-network can be efficiently learned with our algorithm. Comprehensive comparison experiments and analyses validate its effectiveness for robust learning with noisy labels, particularly in the presence of open-set noise. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Computer Vision is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
Text:
  Availability: 1
Header DbId: egs
DbLabel: Engineering Source
An: 182410329
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Variational Rectification Inference for Learning with Noisy Labels.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Sun%2C+Haoliang%22">Sun, Haoliang</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> haolsun@sdu.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Wei%2C+Qi%22">Wei, Qi</searchLink><relatesTo>2</relatesTo> (AUTHOR)<i> 1998v7@gmail.com</i><br /><searchLink fieldCode="AR" term="%22Feng%2C+Lei%22">Feng, Lei</searchLink><relatesTo>3</relatesTo> (AUTHOR)<i> feng_lei@sutd.edu.sg</i><br /><searchLink fieldCode="AR" term="%22Hu%2C+Yupeng%22">Hu, Yupeng</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> huyupeng@sdu.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Liu%2C+Fan%22">Liu, Fan</searchLink><relatesTo>4</relatesTo> (AUTHOR)<i> liufancs@gmail.com</i><br /><searchLink fieldCode="AR" term="%22Fan%2C+Hehe%22">Fan, Hehe</searchLink><relatesTo>5</relatesTo> (AUTHOR)<i> hehefan@zju.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Yin%2C+Yilong%22">Yin, Yilong</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> ylyin@sdu.edu.cn</i>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22International+Journal+of+Computer+Vision%22">International Journal of Computer Vision</searchLink>. Feb2025, Vol. 133 Issue 2, p652-671. 20p.
– Name: Subject
  Label: Subjects
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Dirac+function%22">Dirac function</searchLink><br /><searchLink fieldCode="DE" term="%22Artificial+intelligence%22">Artificial intelligence</searchLink><br /><searchLink fieldCode="DE" term="%22Machine+learning%22">Machine learning</searchLink><br /><searchLink fieldCode="DE" term="%22Bilevel+programming%22">Bilevel programming</searchLink><br /><searchLink fieldCode="DE" term="%22Latent+variables%22">Latent variables</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Label noise has been broadly observed in real-world datasets. To mitigate the negative impact of overfitting to label noise for deep models, effective strategies (e.g., re-weighting, or loss rectification) have been broadly applied in prevailing approaches, which have been generally learned under the meta-learning scenario. Despite the robustness of noise achieved by the probabilistic meta-learning models, they usually suffer from model collapse that degenerates generalization performance. In this paper, we propose variational rectification inference (VRI) to formulate the adaptive rectification for loss functions as an amortized variational inference problem and derive the evidence lower bound under the meta-learning framework. Specifically, VRI is constructed as a hierarchical Bayes by treating the rectifying vector as a latent variable, which can rectify the loss of the noisy sample with the extra randomness regularization and is, therefore, more robust to label noise. To achieve the inference of the rectifying vector, we approximate its conditional posterior with an amortization meta-network. By introducing the variational term in VRI, the conditional posterior is estimated accurately and avoids collapsing to a Dirac delta function, which can significantly improve the generalization performance. The elaborated meta-network and prior network adhere to the smoothness assumption, enabling the generation of reliable rectification vectors. Given a set of clean meta-data, VRI can be efficiently meta-learned within the bi-level optimization programming. Besides, theoretical analysis guarantees that the meta-network can be efficiently learned with our algorithm. Comprehensive comparison experiments and analyses validate its effectiveness for robust learning with noisy labels, particularly in the presence of open-set noise. [ABSTRACT FROM AUTHOR]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: <i>Copyright of International Journal of Computer Vision is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=182410329
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1007/s11263-024-02205-5
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 20
        StartPage: 652
    Subjects:
      – SubjectFull: Dirac function
        Type: general
      – SubjectFull: Artificial intelligence
        Type: general
      – SubjectFull: Machine learning
        Type: general
      – SubjectFull: Bilevel programming
        Type: general
      – SubjectFull: Latent variables
        Type: general
    Titles:
      – TitleFull: Variational Rectification Inference for Learning with Noisy Labels.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Sun, Haoliang
      – PersonEntity:
          Name:
            NameFull: Wei, Qi
      – PersonEntity:
          Name:
            NameFull: Feng, Lei
      – PersonEntity:
          Name:
            NameFull: Hu, Yupeng
      – PersonEntity:
          Name:
            NameFull: Liu, Fan
      – PersonEntity:
          Name:
            NameFull: Fan, Hehe
      – PersonEntity:
          Name:
            NameFull: Yin, Yilong
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 02
              Text: Feb2025
              Type: published
              Y: 2025
          Identifiers:
            – Type: issn-print
              Value: 09205691
          Numbering:
            – Type: volume
              Value: 133
            – Type: issue
              Value: 2
          Titles:
            – TitleFull: International Journal of Computer Vision
              Type: main
ResultId 1