Constructing a multi-class classifier using one-against-one approach with different binary classifiers

Seokho Kang, Sungzoon Cho, Pilsung Kang

Research output: Contribution to journalArticle

32 Citations (Scopus)

Abstract

For the one-against-one approach, all the binary classifiers that form a one-against-one classifier should be sufficiently competent. If some of the classifiers are not competent, the consequences might be invalid classification results. To address the problem, we propose diversified one-against-one (DOAO) method that seeks to find the best classification algorithm for each class pair when applying the one-against-one approach to multi-class classification problems. Applying the proposed method makes various classification algorithms to complement each other. Since the best classification algorithm for each class pair is different, the proposed method can obtain improved classification results. Experimental results show that the proposed method outperforms other one-against-one based methods.

Original languageEnglish
Pages (from-to)677-682
Number of pages6
JournalNeurocomputing
Volume149
Issue numberPB
DOIs
Publication statusPublished - 2015 Feb 3
Externally publishedYes

Fingerprint

Classifiers

Keywords

  • Diversified one-against-one
  • Ensemble
  • Multi-class classification
  • One-against-one

ASJC Scopus subject areas

  • Computer Science Applications
  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this

Constructing a multi-class classifier using one-against-one approach with different binary classifiers. / Kang, Seokho; Cho, Sungzoon; Kang, Pilsung.

In: Neurocomputing, Vol. 149, No. PB, 03.02.2015, p. 677-682.

Research output: Contribution to journalArticle

@article{37b2c372562543fc9b6ade7af08714c0,
title = "Constructing a multi-class classifier using one-against-one approach with different binary classifiers",
abstract = "For the one-against-one approach, all the binary classifiers that form a one-against-one classifier should be sufficiently competent. If some of the classifiers are not competent, the consequences might be invalid classification results. To address the problem, we propose diversified one-against-one (DOAO) method that seeks to find the best classification algorithm for each class pair when applying the one-against-one approach to multi-class classification problems. Applying the proposed method makes various classification algorithms to complement each other. Since the best classification algorithm for each class pair is different, the proposed method can obtain improved classification results. Experimental results show that the proposed method outperforms other one-against-one based methods.",
keywords = "Diversified one-against-one, Ensemble, Multi-class classification, One-against-one",
author = "Seokho Kang and Sungzoon Cho and Pilsung Kang",
year = "2015",
month = "2",
day = "3",
doi = "10.1016/j.neucom.2014.08.006",
language = "English",
volume = "149",
pages = "677--682",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",
number = "PB",

}

TY - JOUR

T1 - Constructing a multi-class classifier using one-against-one approach with different binary classifiers

AU - Kang, Seokho

AU - Cho, Sungzoon

AU - Kang, Pilsung

PY - 2015/2/3

Y1 - 2015/2/3

N2 - For the one-against-one approach, all the binary classifiers that form a one-against-one classifier should be sufficiently competent. If some of the classifiers are not competent, the consequences might be invalid classification results. To address the problem, we propose diversified one-against-one (DOAO) method that seeks to find the best classification algorithm for each class pair when applying the one-against-one approach to multi-class classification problems. Applying the proposed method makes various classification algorithms to complement each other. Since the best classification algorithm for each class pair is different, the proposed method can obtain improved classification results. Experimental results show that the proposed method outperforms other one-against-one based methods.

AB - For the one-against-one approach, all the binary classifiers that form a one-against-one classifier should be sufficiently competent. If some of the classifiers are not competent, the consequences might be invalid classification results. To address the problem, we propose diversified one-against-one (DOAO) method that seeks to find the best classification algorithm for each class pair when applying the one-against-one approach to multi-class classification problems. Applying the proposed method makes various classification algorithms to complement each other. Since the best classification algorithm for each class pair is different, the proposed method can obtain improved classification results. Experimental results show that the proposed method outperforms other one-against-one based methods.

KW - Diversified one-against-one

KW - Ensemble

KW - Multi-class classification

KW - One-against-one

UR - http://www.scopus.com/inward/record.url?scp=84922450521&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84922450521&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2014.08.006

DO - 10.1016/j.neucom.2014.08.006

M3 - Article

AN - SCOPUS:84922450521

VL - 149

SP - 677

EP - 682

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

IS - PB

ER -