Combining multiple classifiers based on third-order dependency

Hee Joong Kang; David Doermann

doi:10.1109/ICDAR.2003.1227621

Combining multiple classifiers based on third-order dependency

Hee Joong Kang
, David Doermann

Department of Computer Science and Engineering

Hansung University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Without an independence assumption, combining multiple classifiers deals with a high order probability distribution composed of classifiers and a class label. Storing and estimating the high order probability distribution is exponentially complex and unmanageable in theoretical analysis, so we rely on an approximation scheme using the dependency. In this paper, as an extension of the second-order dependency approach, the probability distribution is optimally approximated by the third-order dependency and multiple classifiers are combined. The proposed method is evaluated on the recognition of unconstrained handwritten numerals from Concordia University and the University of California, Irvine. Experimental results support the proposed method as a promising approach.

Original language	English
Title of host publication	Proceedings - 7th International Conference on Document Analysis and Recognition, ICDAR 2003
Publisher	IEEE Computer Society
Pages	21-25
Number of pages	5
ISBN (Electronic)	0769519601
DOIs	https://doi.org/10.1109/ICDAR.2003.1227621
State	Published - 2003
Event	7th International Conference on Document Analysis and Recognition, ICDAR 2003 - Edinburgh, United Kingdom Duration: Aug 3 2003 → Aug 6 2003

Publication series

Name	Proceedings of the International Conference on Document Analysis and Recognition, ICDAR
Volume	2003-January
ISSN (Print)	1520-5363

Conference

Conference	7th International Conference on Document Analysis and Recognition, ICDAR 2003
Country/Territory	United Kingdom
City	Edinburgh
Period	08/3/03 → 08/6/03

Access to Document

10.1109/ICDAR.2003.1227621

Cite this

Kang, H. J., & Doermann, D. (2003). Combining multiple classifiers based on third-order dependency. In Proceedings - 7th International Conference on Document Analysis and Recognition, ICDAR 2003 (pp. 21-25). Article 1227621 (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR; Vol. 2003-January). IEEE Computer Society. https://doi.org/10.1109/ICDAR.2003.1227621

@inproceedings{97aed62cac5c47e2acba09b1abdf2768,

title = "Combining multiple classifiers based on third-order dependency",

abstract = "Without an independence assumption, combining multiple classifiers deals with a high order probability distribution composed of classifiers and a class label. Storing and estimating the high order probability distribution is exponentially complex and unmanageable in theoretical analysis, so we rely on an approximation scheme using the dependency. In this paper, as an extension of the second-order dependency approach, the probability distribution is optimally approximated by the third-order dependency and multiple classifiers are combined. The proposed method is evaluated on the recognition of unconstrained handwritten numerals from Concordia University and the University of California, Irvine. Experimental results support the proposed method as a promising approach.",

author = "Kang, \{Hee Joong\} and David Doermann",

note = "Publisher Copyright: {\textcopyright} 2003 IEEE.; 7th International Conference on Document Analysis and Recognition, ICDAR 2003 ; Conference date: 03-08-2003 Through 06-08-2003",

year = "2003",

doi = "10.1109/ICDAR.2003.1227621",

language = "English",

series = "Proceedings of the International Conference on Document Analysis and Recognition, ICDAR",

publisher = "IEEE Computer Society",

pages = "21--25",

booktitle = "Proceedings - 7th International Conference on Document Analysis and Recognition, ICDAR 2003",

address = "United States",

}

Kang, HJ & Doermann, D 2003, Combining multiple classifiers based on third-order dependency. in Proceedings - 7th International Conference on Document Analysis and Recognition, ICDAR 2003., 1227621, Proceedings of the International Conference on Document Analysis and Recognition, ICDAR, vol. 2003-January, IEEE Computer Society, pp. 21-25, 7th International Conference on Document Analysis and Recognition, ICDAR 2003, Edinburgh, United Kingdom, 08/3/03. https://doi.org/10.1109/ICDAR.2003.1227621

Combining multiple classifiers based on third-order dependency. / Kang, Hee Joong; Doermann, David.
Proceedings - 7th International Conference on Document Analysis and Recognition, ICDAR 2003. IEEE Computer Society, 2003. p. 21-25 1227621 (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR; Vol. 2003-January).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Combining multiple classifiers based on third-order dependency

AU - Kang, Hee Joong

AU - Doermann, David

PY - 2003

Y1 - 2003

N2 - Without an independence assumption, combining multiple classifiers deals with a high order probability distribution composed of classifiers and a class label. Storing and estimating the high order probability distribution is exponentially complex and unmanageable in theoretical analysis, so we rely on an approximation scheme using the dependency. In this paper, as an extension of the second-order dependency approach, the probability distribution is optimally approximated by the third-order dependency and multiple classifiers are combined. The proposed method is evaluated on the recognition of unconstrained handwritten numerals from Concordia University and the University of California, Irvine. Experimental results support the proposed method as a promising approach.

AB - Without an independence assumption, combining multiple classifiers deals with a high order probability distribution composed of classifiers and a class label. Storing and estimating the high order probability distribution is exponentially complex and unmanageable in theoretical analysis, so we rely on an approximation scheme using the dependency. In this paper, as an extension of the second-order dependency approach, the probability distribution is optimally approximated by the third-order dependency and multiple classifiers are combined. The proposed method is evaluated on the recognition of unconstrained handwritten numerals from Concordia University and the University of California, Irvine. Experimental results support the proposed method as a promising approach.

UR - https://www.scopus.com/pages/publications/84945955122

U2 - 10.1109/ICDAR.2003.1227621

DO - 10.1109/ICDAR.2003.1227621

M3 - Conference contribution

AN - SCOPUS:84945955122

T3 - Proceedings of the International Conference on Document Analysis and Recognition, ICDAR

SP - 21

EP - 25

BT - Proceedings - 7th International Conference on Document Analysis and Recognition, ICDAR 2003

PB - IEEE Computer Society

T2 - 7th International Conference on Document Analysis and Recognition, ICDAR 2003

Y2 - 3 August 2003 through 6 August 2003

ER -

Combining multiple classifiers based on third-order dependency

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this