TU Darmstadt / ULB / TUbiblio

Pairwise Learning of Multilabel Classifications with Perceptrons

Loza Mencía, Eneldo ; Fürnkranz, Johannes (2007)
Pairwise Learning of Multilabel Classifications with Perceptrons.
Report, Bibliographie

Kurzbeschreibung (Abstract)

Multiclass Multilabel Perceptrons (MMP) are an efficient incremental algorithm for training a team of perceptrons for a multilabel prediction task. The key idea is to train one binary classifier per label, as is typically done for addressing multilabel problems, but to make the training signal dependent on the performance of the whole ensemble. In this paper, we propose an alternative approach that is based on a pairwise approach, i.e., we incrementally train a perceptron for each pair of classes. An evaluation on the Reuters 2000 (RCV1) data shows that our multilabel pairwise perceptron (MLPP) algorithm yields substantial improvements over MMP in terms of ranking quality and overfitting resistance, while maintaining its efficiency. Despite the quadratic increase in the number of perceptrons that have to be trained, the increase in computational complexity is bounded by the average number of labels per training example.

Typ des Eintrags: Report
Erschienen: 2007
Autor(en): Loza Mencía, Eneldo ; Fürnkranz, Johannes
Art des Eintrags: Bibliographie
Titel: Pairwise Learning of Multilabel Classifications with Perceptrons
Sprache: Englisch
Publikationsjahr: 2007
URL / URN: http://www.ke.informatik.tu-darmstadt.de/publications/report...
Kurzbeschreibung (Abstract):

Multiclass Multilabel Perceptrons (MMP) are an efficient incremental algorithm for training a team of perceptrons for a multilabel prediction task. The key idea is to train one binary classifier per label, as is typically done for addressing multilabel problems, but to make the training signal dependent on the performance of the whole ensemble. In this paper, we propose an alternative approach that is based on a pairwise approach, i.e., we incrementally train a perceptron for each pair of classes. An evaluation on the Reuters 2000 (RCV1) data shows that our multilabel pairwise perceptron (MLPP) algorithm yields substantial improvements over MMP in terms of ranking quality and overfitting resistance, while maintaining its efficiency. Despite the quadratic increase in the number of perceptrons that have to be trained, the increase in computational complexity is bounded by the average number of labels per training example.

ID-Nummer: TUD-KE-2007-05
Fachbereich(e)/-gebiet(e): 20 Fachbereich Informatik
20 Fachbereich Informatik > Knowledge Engineering
Hinterlegungsdatum: 24 Jun 2011 15:25
Letzte Änderung: 26 Aug 2018 21:26
PPN:
Export:
Suche nach Titel in: TUfind oder in Google
Frage zum Eintrag Frage zum Eintrag

Optionen (nur für Redakteure)
Redaktionelle Details anzeigen Redaktionelle Details anzeigen