EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Learning label smoothing for text classification

Han Ren iD, Yajie Zhao iD, Yong Zhang, Wei Sun

PeerJ Computer Science · 23.04.2024 · 10 · e2005

DOI: 10.7717/peerj-cs.2005 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PeerJ Computer Science, online erschienen am 23.04.2024. Verfasst von Han Ren und 3 weiteren Autor:innen. Beteiligte Einrichtung: Laboratory of Language Engineering and Computing, Guangdong University of Foreign Studies, Guangzhou, China. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Philosophy and Social Sciences of the Ministry of Education, Research Fund of National Language Commission, Guangdong Education Department Project Foundation.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Training with soft labels instead of hard labels can effectively improve the robustness and generalization of deep learning models. Label smoothing often provides uniformly distributed soft labels during the training process, whereas it does not take the semantic difference of labels into account. This article introduces discrimination-aware label smoothing, an adaptive label smoothing approach that learns appropriate distributions of labels for iterative optimization objectives. In this approach, positive and negative samples are employed to provide experience from both sides, and the performances of regularization and model calibration are improved through an iterative learning method. Experiments on five text classification datasets demonstrate the effectiveness of the proposed method.

Schlagwörter

Stichwörter der Autor:innen

Excessive regularizationLabel smoothingNeural networkSoft labelText classification

Autor:innen und Institutionen

  • Laboratory of Language Engineering and Computing, Guangdong University of Foreign Studies, Guangzhou, China.
  • Laboratory of Language and Artificial Intelligence, Guangdong University of Foreign Studies, Guangzhou, China.
  • School of Information Science and Technology, Guangdong University of Foreign Studies, Guangzhou, China.
  • School of Computer Science, Central China Normal University, Wuhan, China.
  • School of Information Science and Technology, Qiong Tai Normal University, Haikou, China.

ORCID-Profile: Han Ren, Yajie Zhao

Förderung

  • Philosophy and Social Sciences of the Ministry of Education
  • Research Fund of National Language Commission
  • Guangdong Education Department Project Foundation
  • Guangdong Philosophy and Social Sciences Foundation
  • Guangdong University of Foreign Studies Project Foundation
  • Guangzhou Science and Technology Project Foundation
  • National Natural Science Foundation of China
  • Hainan Natural Science Foundation

Zitieren

APA

Ren, H., Zhao, Y., Zhang, Y., & Sun, W. (2024). Learning label smoothing for text classification. PeerJ Computer Science, 10, e2005. https://doi.org/10.7717/peerj-cs.2005

Vancouver

Ren H, Zhao Y, Zhang Y, Sun W. Learning label smoothing for text classification. PeerJ Computer Science. 2024;10:e2005. doi:10.7717/peerj-cs.2005

BibTeX
@article{Ren2024,
  author = {Ren, Han and Zhao, Yajie and Zhang, Yong and Sun, Wei},
  title = {Learning label smoothing for text classification},
  journal = {PeerJ Computer Science},
  year = {2024},
  volume = {10},
  pages = {e2005},
  doi = {10.7717/peerj-cs.2005}
}
Kostenlos starten →