Pattern classification based on local learning

Jing Peng, Bir Bhanu

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

Local learning methods approximate a target function (a posteriori probability) by partitioning the input space into a set of local regions, and modeling a simple input-output relationship in each one. In order for local learning to be effective for pattern classification in high dimensional settings, regions must be chosen judiciously to minimize bias. This paper presents a novel region partitioning criterion that attempts to minimize bias by capturing differential relevance in input variables in an efficient way. The efficacy of the method is validated using a variety of real and simulated data.

Original languageEnglish
Title of host publicationAdvances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings
EditorsAdnan Amin, Dov Dori, Pavel Pudil, Herbert Freeman
PublisherSpringer Verlag
Pages882-889
Number of pages8
ISBN (Print)3540648585, 9783540648581
StatePublished - 1 Jan 1998
Event7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998 - Sydney, Australia
Duration: 11 Aug 199813 Aug 1998

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1451
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998
CountryAustralia
CitySydney
Period11/08/9813/08/98

Fingerprint

Pattern Classification
Pattern recognition
Partitioning
Minimise
Efficacy
High-dimensional
Target
Output
Modeling
Learning

Cite this

Peng, J., & Bhanu, B. (1998). Pattern classification based on local learning. In A. Amin, D. Dori, P. Pudil, & H. Freeman (Eds.), Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings (pp. 882-889). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1451). Springer Verlag.
Peng, Jing ; Bhanu, Bir. / Pattern classification based on local learning. Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. editor / Adnan Amin ; Dov Dori ; Pavel Pudil ; Herbert Freeman. Springer Verlag, 1998. pp. 882-889 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{88033a4d28bd47b3a67f9c83b13a0cb9,
title = "Pattern classification based on local learning",
abstract = "Local learning methods approximate a target function (a posteriori probability) by partitioning the input space into a set of local regions, and modeling a simple input-output relationship in each one. In order for local learning to be effective for pattern classification in high dimensional settings, regions must be chosen judiciously to minimize bias. This paper presents a novel region partitioning criterion that attempts to minimize bias by capturing differential relevance in input variables in an efficient way. The efficacy of the method is validated using a variety of real and simulated data.",
author = "Jing Peng and Bir Bhanu",
year = "1998",
month = "1",
day = "1",
language = "English",
isbn = "3540648585",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "882--889",
editor = "Adnan Amin and Dov Dori and Pavel Pudil and Herbert Freeman",
booktitle = "Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings",

}

Peng, J & Bhanu, B 1998, Pattern classification based on local learning. in A Amin, D Dori, P Pudil & H Freeman (eds), Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1451, Springer Verlag, pp. 882-889, 7th Joint IAPR International Workshop on Structural and Syntactic Pattern Recognition, SSPR 1998 and 2nd International Workshop on Statistical Techniques in Pattern Recognition, SPR 1998, Sydney, Australia, 11/08/98.

Pattern classification based on local learning. / Peng, Jing; Bhanu, Bir.

Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. ed. / Adnan Amin; Dov Dori; Pavel Pudil; Herbert Freeman. Springer Verlag, 1998. p. 882-889 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1451).

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

TY - GEN

T1 - Pattern classification based on local learning

AU - Peng, Jing

AU - Bhanu, Bir

PY - 1998/1/1

Y1 - 1998/1/1

N2 - Local learning methods approximate a target function (a posteriori probability) by partitioning the input space into a set of local regions, and modeling a simple input-output relationship in each one. In order for local learning to be effective for pattern classification in high dimensional settings, regions must be chosen judiciously to minimize bias. This paper presents a novel region partitioning criterion that attempts to minimize bias by capturing differential relevance in input variables in an efficient way. The efficacy of the method is validated using a variety of real and simulated data.

AB - Local learning methods approximate a target function (a posteriori probability) by partitioning the input space into a set of local regions, and modeling a simple input-output relationship in each one. In order for local learning to be effective for pattern classification in high dimensional settings, regions must be chosen judiciously to minimize bias. This paper presents a novel region partitioning criterion that attempts to minimize bias by capturing differential relevance in input variables in an efficient way. The efficacy of the method is validated using a variety of real and simulated data.

UR - http://www.scopus.com/inward/record.url?scp=84947777485&partnerID=8YFLogxK

M3 - Conference contribution

SN - 3540648585

SN - 9783540648581

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

SP - 882

EP - 889

BT - Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings

A2 - Amin, Adnan

A2 - Dori, Dov

A2 - Pudil, Pavel

A2 - Freeman, Herbert

PB - Springer Verlag

ER -

Peng J, Bhanu B. Pattern classification based on local learning. In Amin A, Dori D, Pudil P, Freeman H, editors, Advances in Pattern Recognition - Joint IAPR International Workshops SSPR 1998 and SPR 1998, Proceedings. Springer Verlag. 1998. p. 882-889. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).