Predicting unseen labels using label hierarchies in large-scale multi-label learning

Jinseok Nam, Eneldo Loza Mencía, Hyun Woo Kim, Johannes Fürnkranz

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

6 Citations (Scopus)

Abstract

An important problem in multi-label classification is to capture label patterns or underlying structures that have an impact on such patterns. One way of learning underlying structures over labels is to project both instances and labels into the same space where an instance and its relevant labels tend to have similar representations. In this paper, we present a novel method to learn a joint space of instances and labels by leveraging a hierarchy of labels. We also present an efficient method for pretraining vector representations of labels, namely label embeddings, from large amounts of label co-occurrence patterns and hierarchical structures of labels. This approach also allows us to make predictions on labels that have not been seen during training. We empirically show that the use of pretrained label embeddings allows us to obtain higher accuracies on unseen labels even when the number of labels are quite large. Our experimental results also demonstrate qualitatively that the proposed method is able to learn regularities among labels by exploiting a label hierarchy as well as label co-occurrences.

Original languageEnglish
Title of host publicationMachine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015, Proceedings
EditorsAnnalisa Appice, João Gama, Vitor Santos Costa, João Gama, Alípio Jorge, Annalisa Appice, Annalisa Appice, Vitor Santos Costa, Alípio Jorge, Annalisa Appice, Pedro Pereira Rodrigues, Pedro Pereira Rodrigues, João Gama, Vitor Santos Costa, Soares Soares, Pedro Pereira Rodrigues, Soares Soares, Soares Soares, João Gama, Soares Soares, Alípio Jorge, Alípio Jorge, Pedro Pereira Rodrigues, Vitor Santos Costa
PublisherSpringer Verlag
Pages102-118
Number of pages17
ISBN (Print)9783319235271, 9783319235271, 9783319235271, 9783319235271
DOIs
Publication statusPublished - 2015 Jan 1
Externally publishedYes
EventEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015 - Porto, Portugal
Duration: 2015 Sep 72015 Sep 11

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9284
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

OtherEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2015
CountryPortugal
CityPorto
Period15/9/715/9/11

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Predicting unseen labels using label hierarchies in large-scale multi-label learning'. Together they form a unique fingerprint.

  • Cite this

    Nam, J., Mencía, E. L., Kim, H. W., & Fürnkranz, J. (2015). Predicting unseen labels using label hierarchies in large-scale multi-label learning. In A. Appice, J. Gama, V. S. Costa, J. Gama, A. Jorge, A. Appice, A. Appice, V. S. Costa, A. Jorge, A. Appice, P. P. Rodrigues, P. P. Rodrigues, J. Gama, V. S. Costa, S. Soares, P. P. Rodrigues, S. Soares, S. Soares, J. Gama, S. Soares, A. Jorge, A. Jorge, P. P. Rodrigues, ... V. S. Costa (Eds.), Machine Learning and Knowledge Discovery in Databases - European Conference, ECML PKDD 2015, Proceedings (pp. 102-118). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9284). Springer Verlag. https://doi.org/10.1007/978-3-319-23528-8_7