Computationally efficient wasserstein loss for structured labels

Ayato Toyokuni, Sho Yokoi, Hisashi Kashima, Makoto Yamada

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The problem of estimating the probability distribution of labels has been widely studied as a label distribution learning (LDL) problem, whose applications include age estimation, emotion analysis, and semantic segmentation. We propose a tree-Wasserstein distance regularized LDL algorithm, focusing on hierarchical text classification tasks. We propose predicting the entire label hierarchy using neural networks, where the similarity between predicted and true labels is measured using the tree-Wasserstein distance. Through experiments using synthetic and real-world datasets, we demonstrate that the proposed method successfully considers the structure of labels during training, and it compares favorably with the Sinkhorn algorithm in terms of computation time and memory usage.

Original languageEnglish
Title of host publicationEACL 2021 - 16th Conference of the European Chapter of the Association for Computational Linguistics, Proceedings of the Student Research Workshop
PublisherAssociation for Computational Linguistics (ACL)
Pages1-7
Number of pages7
ISBN (Electronic)9781954085046
Publication statusPublished - 2021
Event16th Conference of the European Chapter of the Associationfor Computational Linguistics: Student Research Workshop, EACL 2021 - Virtual, Online
Duration: 2021 Apr 192021 Apr 23

Publication series

NameEACL 2021 - 16th Conference of the European Chapter of the Association for Computational Linguistics, Proceedings of the Student Research Workshop

Conference

Conference16th Conference of the European Chapter of the Associationfor Computational Linguistics: Student Research Workshop, EACL 2021
CityVirtual, Online
Period21/4/1921/4/23

ASJC Scopus subject areas

  • Software
  • Computational Theory and Mathematics
  • Linguistics and Language

Fingerprint

Dive into the research topics of 'Computationally efficient wasserstein loss for structured labels'. Together they form a unique fingerprint.

Cite this