Candidate elimination criteria for Lazy Bayesian Rules

Geoffrey Ian Webb

doi:10.1007/3-540-45656-2_47

Candidate elimination criteria for Lazy Bayesian Rules

Geoffrey Ian Webb

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

11 Citations (Scopus)

Abstract

Lazy Bayesian Rules modifies naive Bayesian classification to undo elements of the harmful attribute independence assumption. It has been shown to provide classification error comparable to boosting decision trees. This paper explores alternatives to the candidate elimination criterion employed within Lazy Bayesian Rules. Improvements over naive Bayes are consistent so long as the candidate elimination criteria ensures there is sufficient data for accurate probability estimation. However, the original candidate elimination criterion is demonstrated to provide better overall error reduction than the use of a minimum data subset size criterion.

Original language	English
Title of host publication	AI 2001: Advances in Artificial Intelligence
Subtitle of host publication	14th Australian Joint Conference on Artificial Intelligence Adelaide, Australia, December 10-14, 2001 Proceedings
Editors	Markus Stumptner, Dan Corbett, Mike Brooks
Place of Publication	Berlin Germany
Publisher	Springer
Pages	545-556
Number of pages	12
ISBN (Print)	3540429603
DOIs	https://doi.org/10.1007/3-540-45656-2_47
Publication status	Published - 2001
Event	Australasian Joint Conference on Artificial Intelligence 2001 - Adelaide, Australia Duration: 10 Dec 2001 → 14 Dec 2001 Conference number: 14th https://link.springer.com/book/10.1007/3-540-45656-2 (Proceedings)

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	2256
ISSN (Print)	0302-9743

Conference

Conference	Australasian Joint Conference on Artificial Intelligence 2001
Abbreviated title	AI 2001
Country/Territory	Australia
City	Adelaide
Period	10/12/01 → 14/12/01
Internet address	https://link.springer.com/book/10.1007/3-540-45656-2 (Proceedings)

Access to Document

10.1007/3-540-45656-2_47

Cite this

Webb, G. I. (2001). Candidate elimination criteria for Lazy Bayesian Rules. In M. Stumptner, D. Corbett, & M. Brooks (Eds.), AI 2001: Advances in Artificial Intelligence: 14th Australian Joint Conference on Artificial Intelligence Adelaide, Australia, December 10-14, 2001 Proceedings (pp. 545-556). (Lecture Notes in Computer Science; Vol. 2256). Springer. https://doi.org/10.1007/3-540-45656-2_47

Webb, Geoffrey Ian. / Candidate elimination criteria for Lazy Bayesian Rules. AI 2001: Advances in Artificial Intelligence: 14th Australian Joint Conference on Artificial Intelligence Adelaide, Australia, December 10-14, 2001 Proceedings. editor / Markus Stumptner ; Dan Corbett ; Mike Brooks. Berlin Germany : Springer, 2001. pp. 545-556 (Lecture Notes in Computer Science).

@inproceedings{89d8a6550ebf4fe3b16291da3eceb4d8,

title = "Candidate elimination criteria for Lazy Bayesian Rules",

abstract = "Lazy Bayesian Rules modifies naive Bayesian classification to undo elements of the harmful attribute independence assumption. It has been shown to provide classification error comparable to boosting decision trees. This paper explores alternatives to the candidate elimination criterion employed within Lazy Bayesian Rules. Improvements over naive Bayes are consistent so long as the candidate elimination criteria ensures there is sufficient data for accurate probability estimation. However, the original candidate elimination criterion is demonstrated to provide better overall error reduction than the use of a minimum data subset size criterion.",

author = "Webb, {Geoffrey Ian}",

year = "2001",

doi = "10.1007/3-540-45656-2_47",

language = "English",

isbn = "3540429603",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "545--556",

editor = "Markus Stumptner and Dan Corbett and Mike Brooks",

booktitle = "AI 2001: Advances in Artificial Intelligence",

note = "Australasian Joint Conference on Artificial Intelligence 2001, AI 2001 ; Conference date: 10-12-2001 Through 14-12-2001",

url = "https://link.springer.com/book/10.1007/3-540-45656-2",

}

Webb, GI 2001, Candidate elimination criteria for Lazy Bayesian Rules. in M Stumptner, D Corbett & M Brooks (eds), AI 2001: Advances in Artificial Intelligence: 14th Australian Joint Conference on Artificial Intelligence Adelaide, Australia, December 10-14, 2001 Proceedings. Lecture Notes in Computer Science, vol. 2256, Springer, Berlin Germany, pp. 545-556, Australasian Joint Conference on Artificial Intelligence 2001, Adelaide, South Australia, Australia, 10/12/01. https://doi.org/10.1007/3-540-45656-2_47

Candidate elimination criteria for Lazy Bayesian Rules. / Webb, Geoffrey Ian.

AI 2001: Advances in Artificial Intelligence: 14th Australian Joint Conference on Artificial Intelligence Adelaide, Australia, December 10-14, 2001 Proceedings. ed. / Markus Stumptner; Dan Corbett; Mike Brooks. Berlin Germany : Springer, 2001. p. 545-556 (Lecture Notes in Computer Science; Vol. 2256).

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

TY - GEN

T1 - Candidate elimination criteria for Lazy Bayesian Rules

AU - Webb, Geoffrey Ian

N1 - Conference code: 14th

PY - 2001

Y1 - 2001

N2 - Lazy Bayesian Rules modifies naive Bayesian classification to undo elements of the harmful attribute independence assumption. It has been shown to provide classification error comparable to boosting decision trees. This paper explores alternatives to the candidate elimination criterion employed within Lazy Bayesian Rules. Improvements over naive Bayes are consistent so long as the candidate elimination criteria ensures there is sufficient data for accurate probability estimation. However, the original candidate elimination criterion is demonstrated to provide better overall error reduction than the use of a minimum data subset size criterion.

AB - Lazy Bayesian Rules modifies naive Bayesian classification to undo elements of the harmful attribute independence assumption. It has been shown to provide classification error comparable to boosting decision trees. This paper explores alternatives to the candidate elimination criterion employed within Lazy Bayesian Rules. Improvements over naive Bayes are consistent so long as the candidate elimination criteria ensures there is sufficient data for accurate probability estimation. However, the original candidate elimination criterion is demonstrated to provide better overall error reduction than the use of a minimum data subset size criterion.

U2 - 10.1007/3-540-45656-2_47

DO - 10.1007/3-540-45656-2_47

M3 - Conference Paper

SN - 3540429603

T3 - Lecture Notes in Computer Science

SP - 545

EP - 556

BT - AI 2001: Advances in Artificial Intelligence

A2 - Stumptner, Markus

A2 - Corbett, Dan

A2 - Brooks, Mike

PB - Springer

CY - Berlin Germany

T2 - Australasian Joint Conference on Artificial Intelligence 2001

Y2 - 10 December 2001 through 14 December 2001

ER -

Webb GI. Candidate elimination criteria for Lazy Bayesian Rules. In Stumptner M, Corbett D, Brooks M, editors, AI 2001: Advances in Artificial Intelligence: 14th Australian Joint Conference on Artificial Intelligence Adelaide, Australia, December 10-14, 2001 Proceedings. Berlin Germany: Springer. 2001. p. 545-556. (Lecture Notes in Computer Science). https://doi.org/10.1007/3-540-45656-2_47