Multi-task algorithms typically use task similarity information as a bias to speed up learning. We argue that, when the classification problem is unbalanced, task dissimilarity information provides a more effective bias, as rare class labels tend to be better separated from the frequent class labels. In particular, we show that a multi-task extension of the label propagation algorithm for graph-based classification works much better on protein function prediction problems when the task relatedness information is represented using a dissimilarity matrix as opposed to a similarity matrix.

Multi-Task Label Propagation with Dissimilarity Measures / M. Frasca, N. Cesa Bianchi - In: BIOKDD'16[s.l] : ACM, 2016 Aug 14. - ISBN 9781450321389. - pp. 1-10 (( Intervento presentato al 15. convegno International Workshop on Data Mining in Bioinformatics tenutosi a San Francisco nel 2016 [10.1145/1235].

Multi-Task Label Propagation with Dissimilarity Measures

M. Frasca;N. Cesa Bianchi
2016

Abstract

Multi-task algorithms typically use task similarity information as a bias to speed up learning. We argue that, when the classification problem is unbalanced, task dissimilarity information provides a more effective bias, as rare class labels tend to be better separated from the frequent class labels. In particular, we show that a multi-task extension of the label propagation algorithm for graph-based classification works much better on protein function prediction problems when the task relatedness information is represented using a dissimilarity matrix as opposed to a similarity matrix.
Multi-task learning; protein function prediction; label propagation algorithm
Settore INF/01 - Informatica
14-ago-2016
http://home.biokdd.org/biokdd16/
Book Part (author)
File in questo prodotto:
File Dimensione Formato  
camera_ready.pdf

accesso riservato

Descrizione: Articolo principale
Tipologia: Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione 215.78 kB
Formato Adobe PDF
215.78 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/501017
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 439
  • ???jsp.display-item.citation.isi??? ND
social impact