Discriminative Methods for Multi-labeled Classification
S. Godbole, and S. Sarawagi. Advances in Knowledge Discovery and Data Mining, (2004)
Abstract
In this paper we present methods of enhancing existing discriminative classifiers for multi-labeled predictions. Discriminative methods like support vector machines perform very well for uni-labeled text classification tasks. Multi-labeled classification is a harder task subject to relatively less attention. In the multi-labeled setting, classes are often related to each other or part of a is-a hierarchy. We present a new technique for combining text features and features indicating relationships between classes, which can be used with any discriminative algorithm. We also present two enhancements to the margin of SVMs for building better models in the presence of overlapping classes. We present results of experiments on real world text benchmark datasets. Our new methods beat accuracy of existing methods with statistically significant improvements.
ER -
%0 Journal Article
%1 Godbole2004Discriminative
%A Godbole, Shantanu
%A Sarawagi, Sunita
%D 2004
%J Advances in Knowledge Discovery and Data Mining
%K 2004 multi-label stacking svm
%P 22--30
%T Discriminative Methods for Multi-labeled Classification
%U http://www.springerlink.com/content/maa4ag38jd3pwrc0
%X In this paper we present methods of enhancing existing discriminative classifiers for multi-labeled predictions. Discriminative methods like support vector machines perform very well for uni-labeled text classification tasks. Multi-labeled classification is a harder task subject to relatively less attention. In the multi-labeled setting, classes are often related to each other or part of a is-a hierarchy. We present a new technique for combining text features and features indicating relationships between classes, which can be used with any discriminative algorithm. We also present two enhancements to the margin of SVMs for building better models in the presence of overlapping classes. We present results of experiments on real world text benchmark datasets. Our new methods beat accuracy of existing methods with statistically significant improvements.
ER -
@article{Godbole2004Discriminative,
abstract = {In this paper we present methods of enhancing existing discriminative classifiers for multi-labeled predictions. Discriminative methods like support vector machines perform very well for uni-labeled text classification tasks. Multi-labeled classification is a harder task subject to relatively less attention. In the multi-labeled setting, classes are often related to each other or part of a is-a hierarchy. We present a new technique for combining text features and features indicating relationships between classes, which can be used with any discriminative algorithm. We also present two enhancements to the margin of SVMs for building better models in the presence of overlapping classes. We present results of experiments on real world text benchmark datasets. Our new methods beat accuracy of existing methods with statistically significant improvements.
ER -},
added-at = {2009-07-08T16:36:43.000+0200},
author = {Godbole, Shantanu and Sarawagi, Sunita},
biburl = {https://www.bibsonomy.org/bibtex/2477686e239894a0fe3349a0f2cf74a2c/zeno},
interhash = {7624432f6d327450b3f2e2e65c85e8c1},
intrahash = {477686e239894a0fe3349a0f2cf74a2c},
journal = {Advances in Knowledge Discovery and Data Mining},
keywords = {2004 multi-label stacking svm},
pages = {22--30},
timestamp = {2009-12-17T17:48:42.000+0100},
title = {Discriminative Methods for Multi-labeled Classification},
url = {http://www.springerlink.com/content/maa4ag38jd3pwrc0},
year = 2004
}