Multi-label classification by label clustering based on covariance

Reem Al-Otaibi, Meelis Kull, Peter Flach

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

103 Downloads (Pure)

Abstract

Multi-label classification is a supervised learning problem that predicts multiple labels simultaneously. One of the key challenges in such tasks is modelling the correlations between multiple labels. LaCovais a decision treemulti-label classifier, that interpolates between two baseline methods: Binary Relevance (BR), which assumes all labels independent; and Label Powerset (LP),which learns the joint label distribution. In this paper we introduce LaCova-CLus that clusters labels into several dependent subsets as an additional splitting criterion. Clusters are obtained locally by identifying the connected components in the thresholded absolute covariance matrix. The proposed algorithm is evaluated and compared to baseline and state-of-the-art approaches. Experimental results show that our method can improve the label exact-match.
Original languageEnglish
Title of host publicationProceedings of the ECMLPKDD 2015 Doctoral Consortium
PublisherAalto University
Pages43-52
Number of pages10
ISBN (Print)9789526064437
Publication statusPublished - 8 Oct 2015
EventEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases - ECML PKDD PhD Consortium - Porto, Portugal, Portugal
Duration: 7 Sep 201511 Sep 2015

Publication series

NameAalto University Publication Series
ISSN (Print)1799-4896

Conference

ConferenceEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases - ECML PKDD PhD Consortium
CountryPortugal
CityPorto, Portugal
Period7/09/1511/09/15

Structured keywords

  • Jean Golding
  • SPHERE

Fingerprint

Dive into the research topics of 'Multi-label classification by label clustering based on covariance'. Together they form a unique fingerprint.

Cite this