TrueSkill-based pairwise coupling for multi-class classification

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

A multi-class classification problem can be solved efficiently via decomposition of the problem into multiple binary classification problems. As a way of such decomposition, we propose a novel pairwise coupling method based on the TrueSkill ranking system. Instead of aggregating all pairwise binary classification results for the final decision, the proposed method keeps track of the ranks of the classes during the successive binary classification procedure. Especially, selection of a binary classifier at a certain step is done in such a way that the multi-class classification decision using the binary classification results up to the step converges to the final one as quickly as possible. Thus, the number of binary classifications can be reduced, which in turn reduces the computational complexity of the whole classification system. Experimental results show that the complexity is reduced significantly for no or minor loss of classification performance.

Original languageEnglish
Title of host publicationArtificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings
Pages213-220
Number of pages8
EditionPART 2
DOIs
Publication statusPublished - 2012 Oct 25
Event22nd International Conference on Artificial Neural Networks, ICANN 2012 - Lausanne, Switzerland
Duration: 2012 Sep 112012 Sep 14

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume7553 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other22nd International Conference on Artificial Neural Networks, ICANN 2012
CountrySwitzerland
CityLausanne
Period12/9/1112/9/14

Fingerprint

Multi-class Classification
Binary Classification
Pairwise
Classification Problems
Decompose
Coupling Method
Minor
Ranking
Computational Complexity
Classifier
Decomposition
Binary
Converge
Experimental Results
Computational complexity
Classifiers

All Science Journal Classification (ASJC) codes

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Lee, J-S. (2012). TrueSkill-based pairwise coupling for multi-class classification. In Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings (PART 2 ed., pp. 213-220). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7553 LNCS, No. PART 2). https://doi.org/10.1007/978-3-642-33266-1_27
Lee, Jong-Seok. / TrueSkill-based pairwise coupling for multi-class classification. Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings. PART 2. ed. 2012. pp. 213-220 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
@inproceedings{4b1c6eb592c04708a8c45b7260fe9d00,
title = "TrueSkill-based pairwise coupling for multi-class classification",
abstract = "A multi-class classification problem can be solved efficiently via decomposition of the problem into multiple binary classification problems. As a way of such decomposition, we propose a novel pairwise coupling method based on the TrueSkill ranking system. Instead of aggregating all pairwise binary classification results for the final decision, the proposed method keeps track of the ranks of the classes during the successive binary classification procedure. Especially, selection of a binary classifier at a certain step is done in such a way that the multi-class classification decision using the binary classification results up to the step converges to the final one as quickly as possible. Thus, the number of binary classifications can be reduced, which in turn reduces the computational complexity of the whole classification system. Experimental results show that the complexity is reduced significantly for no or minor loss of classification performance.",
author = "Jong-Seok Lee",
year = "2012",
month = "10",
day = "25",
doi = "10.1007/978-3-642-33266-1_27",
language = "English",
isbn = "9783642332654",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 2",
pages = "213--220",
booktitle = "Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings",
edition = "PART 2",

}

Lee, J-S 2012, TrueSkill-based pairwise coupling for multi-class classification. in Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 7553 LNCS, pp. 213-220, 22nd International Conference on Artificial Neural Networks, ICANN 2012, Lausanne, Switzerland, 12/9/11. https://doi.org/10.1007/978-3-642-33266-1_27

TrueSkill-based pairwise coupling for multi-class classification. / Lee, Jong-Seok.

Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings. PART 2. ed. 2012. p. 213-220 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7553 LNCS, No. PART 2).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - TrueSkill-based pairwise coupling for multi-class classification

AU - Lee, Jong-Seok

PY - 2012/10/25

Y1 - 2012/10/25

N2 - A multi-class classification problem can be solved efficiently via decomposition of the problem into multiple binary classification problems. As a way of such decomposition, we propose a novel pairwise coupling method based on the TrueSkill ranking system. Instead of aggregating all pairwise binary classification results for the final decision, the proposed method keeps track of the ranks of the classes during the successive binary classification procedure. Especially, selection of a binary classifier at a certain step is done in such a way that the multi-class classification decision using the binary classification results up to the step converges to the final one as quickly as possible. Thus, the number of binary classifications can be reduced, which in turn reduces the computational complexity of the whole classification system. Experimental results show that the complexity is reduced significantly for no or minor loss of classification performance.

AB - A multi-class classification problem can be solved efficiently via decomposition of the problem into multiple binary classification problems. As a way of such decomposition, we propose a novel pairwise coupling method based on the TrueSkill ranking system. Instead of aggregating all pairwise binary classification results for the final decision, the proposed method keeps track of the ranks of the classes during the successive binary classification procedure. Especially, selection of a binary classifier at a certain step is done in such a way that the multi-class classification decision using the binary classification results up to the step converges to the final one as quickly as possible. Thus, the number of binary classifications can be reduced, which in turn reduces the computational complexity of the whole classification system. Experimental results show that the complexity is reduced significantly for no or minor loss of classification performance.

UR - http://www.scopus.com/inward/record.url?scp=84867669578&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867669578&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-33266-1_27

DO - 10.1007/978-3-642-33266-1_27

M3 - Conference contribution

SN - 9783642332654

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 213

EP - 220

BT - Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings

ER -

Lee J-S. TrueSkill-based pairwise coupling for multi-class classification. In Artificial Neural Networks and Machine Learning, ICANN 2012 - 22nd International Conference on Artificial Neural Networks, Proceedings. PART 2 ed. 2012. p. 213-220. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). https://doi.org/10.1007/978-3-642-33266-1_27