Effective and Efficient Multilabel Classification in Domains with Large Number of Labels

G. Tsoumakas, I. Katakis, I. Vlahavas, “Effective and Efficient Multilabel Classification in Domains with Large Number of Labels”, Proc. ECML/PKDD 2008 Workshop on Mining Multidimensional Data (MMD'08), Antwerp, Belgium, 2008.

Proc. ECML/PKDD 2008 Workshop on Mining Multidimensional Data (MMD'08), Antwerp, Belgium, 2008.

This paper contributes a novel algorithm for effective and computationally efficient multilabel classification in domains with large label sets L. The HOMER algorithm constructs a Hierarchy Of Multilabel classifiERs, each one dealing with a much smaller set of labels compared to L and a more balanced example distribution. This leads to improved predictive performance along with linear training and logarithmic testing complexities with respect to |L|. Label distribution from parent to children nodes is achieved via a new balanced clustering algorithm, called balanced k means.