Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Deep Learning for Extreme Multi-label Text Classification

Deep Learning for Extreme Multi-label Text Classification Session 1C: Document Representation and Content Analysis 1 SIGIR ™17, August 7-11, 2017, Shinjuku, Tokyo, Japan Deep Learning for Extreme Multi-label Text Classification Jingzhou Liu Wei-Cheng Chang Carnegie Mellon University liujingzhou@cs.cmu.edu Carnegie Mellon University wchang2@andrew.cmu.edu Yuexin Wu Yiming Yang Carnegie Mellon University yuexinw@andrew.cmu.edu Carnegie Mellon University yiming@cs.cmu.edu extremely large scale presents open challenges for machine learning research. Multi-label classification is fundamentally different from the traditional binary or multi-class classification problems which have been intensively studied in the machine learning literature. Binary classifiers treat class labels as independent target variables, which is clearly sub-optimal for multi-label classification as the dependencies among class labels cannot be leveraged. Multi-class classifiers rely on the mutually exclusive assumption about class labels (i.e., one document should have one and only one class label), which is wrong in multi-label settings. Addressing the limitations of those traditional classification methods by explicitly modeling the dependencies or correlations among class labels has been the major focus of multi-label classification research [7, 11, 13, 15, 42, 48]; however, scalable solutions for problems with hundreds of thousands or even millions of labels have become available only in the past few years[5, 36]. Part of the difficulty in solving XMTC problems http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Deep Learning for Extreme Multi-label Text Classification

Association for Computing Machinery — Aug 7, 2017

Loading next page...
/lp/association-for-computing-machinery/deep-learning-for-extreme-multi-label-text-classification-mWcWFvgffK

References (53)

Datasource
Association for Computing Machinery
Copyright
Copyright © 2017 by ACM Inc.
ISBN
978-1-4503-5022-8
doi
10.1145/3077136.3080834
Publisher site
See Article on Publisher Site

Abstract

Session 1C: Document Representation and Content Analysis 1 SIGIR ™17, August 7-11, 2017, Shinjuku, Tokyo, Japan Deep Learning for Extreme Multi-label Text Classification Jingzhou Liu Wei-Cheng Chang Carnegie Mellon University liujingzhou@cs.cmu.edu Carnegie Mellon University wchang2@andrew.cmu.edu Yuexin Wu Yiming Yang Carnegie Mellon University yuexinw@andrew.cmu.edu Carnegie Mellon University yiming@cs.cmu.edu extremely large scale presents open challenges for machine learning research. Multi-label classification is fundamentally different from the traditional binary or multi-class classification problems which have been intensively studied in the machine learning literature. Binary classifiers treat class labels as independent target variables, which is clearly sub-optimal for multi-label classification as the dependencies among class labels cannot be leveraged. Multi-class classifiers rely on the mutually exclusive assumption about class labels (i.e., one document should have one and only one class label), which is wrong in multi-label settings. Addressing the limitations of those traditional classification methods by explicitly modeling the dependencies or correlations among class labels has been the major focus of multi-label classification research [7, 11, 13, 15, 42, 48]; however, scalable solutions for problems with hundreds of thousands or even millions of labels have become available only in the past few years[5, 36]. Part of the difficulty in solving XMTC problems

There are no references for this article.