TU Darmstadt / ULB / TUbiblio

Graded Multilabel Classification by Pairwise Comparisons

Brinker, Christian and Loza Mencía, Eneldo and Fürnkranz, Johannes (2014):
Graded Multilabel Classification by Pairwise Comparisons.
Shenzhen, China, Curran Associates, IEEE, In: 2014 IEEE International Conference on Data Mining (ICDM 2014), Shenzhen, China, ISSN 1550-4786,
ISBN 978-1-4799-4303-6,
[Online-Edition: http://www.ke.tu-darmstadt.de/publications/papers/ICDM14grad...],
[Conference or Workshop Item]

Abstract

The task in multilabel classification is to predict for a given set of labels whether each individual label should be attached to an instance or not. Graded multilabel classification generalizes this setting by allowing to specify for each label a degree of membership on an ordinal scale. This setting can be frequently found in practice, for example when movies or books are assessed on a one-to-five star rating in multiple categories. In this paper, we propose to reformulate the problem in terms of preferences between the labels and their scales, which then be tackled by learning from pairwise comparisons. We present three different approaches which make use of this decomposition and show on three datasets that we are able to outperform baseline approaches. In particular, we show that our solution, which is able to model pairwise preferences across multiple scales, outperforms a straight-forward approach which considers the problem as a set of independent ordinal regression tasks.

Item Type: Conference or Workshop Item
Erschienen: 2014
Creators: Brinker, Christian and Loza Mencía, Eneldo and Fürnkranz, Johannes
Title: Graded Multilabel Classification by Pairwise Comparisons
Language: English
Abstract:

The task in multilabel classification is to predict for a given set of labels whether each individual label should be attached to an instance or not. Graded multilabel classification generalizes this setting by allowing to specify for each label a degree of membership on an ordinal scale. This setting can be frequently found in practice, for example when movies or books are assessed on a one-to-five star rating in multiple categories. In this paper, we propose to reformulate the problem in terms of preferences between the labels and their scales, which then be tackled by learning from pairwise comparisons. We present three different approaches which make use of this decomposition and show on three datasets that we are able to outperform baseline approaches. In particular, we show that our solution, which is able to model pairwise preferences across multiple scales, outperforms a straight-forward approach which considers the problem as a set of independent ordinal regression tasks.

Place of Publication: Shenzhen, China
Publisher: Curran Associates, IEEE
ISBN: 978-1-4799-4303-6
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Knowl­edge En­gi­neer­ing
Event Title: 2014 IEEE International Conference on Data Mining (ICDM 2014)
Event Location: Shenzhen, China
Date Deposited: 26 Nov 2015 08:26
Official URL: http://www.ke.tu-darmstadt.de/publications/papers/ICDM14grad...
Identification Number: doi:10.1109/ICDM.2014.102
Export:

Optionen (nur für Redakteure)

View Item View Item