Selected tree classifier combination based on both accuracy and error diversity

H. W. Shin, So Young Sohn

Research output: Contribution to journalArticle

29 Citations (Scopus)

Abstract

This paper proposes a method for combining multiple tree classifiers based on both classifier ensemble (bagging) and dynamic classifier selection schemes (DCS). The proposed method is composed of the following procedures: (1) building individual tree classifiers based on bootstrap samples; (2) calculating the distance between all possible two trees; (3) clustering the trees based on single linkage clustering; (4) selecting two clusters by local region in terms of accuracy and error diversity; and (5) voting the results of tree classifiers selected in the two clusters. Empirical evaluation using publicly available data sets confirms the superiority of our proposed approach over other classifier combining methods.

Original languageEnglish
Pages (from-to)191-197
Number of pages7
JournalPattern Recognition
Volume38
Issue number2
DOIs
Publication statusPublished - 2005 Jan 1

Fingerprint

Classifiers

All Science Journal Classification (ASJC) codes

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

@article{9fcf60361d504e73b5e14d7ffdb23c86,
title = "Selected tree classifier combination based on both accuracy and error diversity",
abstract = "This paper proposes a method for combining multiple tree classifiers based on both classifier ensemble (bagging) and dynamic classifier selection schemes (DCS). The proposed method is composed of the following procedures: (1) building individual tree classifiers based on bootstrap samples; (2) calculating the distance between all possible two trees; (3) clustering the trees based on single linkage clustering; (4) selecting two clusters by local region in terms of accuracy and error diversity; and (5) voting the results of tree classifiers selected in the two clusters. Empirical evaluation using publicly available data sets confirms the superiority of our proposed approach over other classifier combining methods.",
author = "Shin, {H. W.} and Sohn, {So Young}",
year = "2005",
month = "1",
day = "1",
doi = "10.1016/S0031-3203(04)00272-9",
language = "English",
volume = "38",
pages = "191--197",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "2",

}

Selected tree classifier combination based on both accuracy and error diversity. / Shin, H. W.; Sohn, So Young.

In: Pattern Recognition, Vol. 38, No. 2, 01.01.2005, p. 191-197.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Selected tree classifier combination based on both accuracy and error diversity

AU - Shin, H. W.

AU - Sohn, So Young

PY - 2005/1/1

Y1 - 2005/1/1

N2 - This paper proposes a method for combining multiple tree classifiers based on both classifier ensemble (bagging) and dynamic classifier selection schemes (DCS). The proposed method is composed of the following procedures: (1) building individual tree classifiers based on bootstrap samples; (2) calculating the distance between all possible two trees; (3) clustering the trees based on single linkage clustering; (4) selecting two clusters by local region in terms of accuracy and error diversity; and (5) voting the results of tree classifiers selected in the two clusters. Empirical evaluation using publicly available data sets confirms the superiority of our proposed approach over other classifier combining methods.

AB - This paper proposes a method for combining multiple tree classifiers based on both classifier ensemble (bagging) and dynamic classifier selection schemes (DCS). The proposed method is composed of the following procedures: (1) building individual tree classifiers based on bootstrap samples; (2) calculating the distance between all possible two trees; (3) clustering the trees based on single linkage clustering; (4) selecting two clusters by local region in terms of accuracy and error diversity; and (5) voting the results of tree classifiers selected in the two clusters. Empirical evaluation using publicly available data sets confirms the superiority of our proposed approach over other classifier combining methods.

UR - http://www.scopus.com/inward/record.url?scp=6444238786&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=6444238786&partnerID=8YFLogxK

U2 - 10.1016/S0031-3203(04)00272-9

DO - 10.1016/S0031-3203(04)00272-9

M3 - Article

AN - SCOPUS:6444238786

VL - 38

SP - 191

EP - 197

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 2

ER -