Canonical Forest

Yu Chuan Chen, Hyejung Ha, Hyunjoong Kim, Hongshik Ahn

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We propose a new classification ensemble method named Canonical Forest. The new method uses canonical linear discriminant analysis (CLDA) and bootstrapping to obtain accurate and diverse classifiers that constitute an ensemble. We note CLDA serves as a linear transformation tool rather than a dimension reduction tool. Since CLDA will find the transformed space that separates the classes farther in distribution, classifiers built on this space will be more accurate than those on the original space. To further facilitate the diversity of the classifiers in an ensemble, CLDA is applied only on a partial feature space for each bootstrapped data. To compare the performance of Canonical Forest and other widely used ensemble methods, we tested them on 29 real or artificial data sets. Canonical Forest performed significantly better in accuracy than other ensemble methods in most data sets. According to the investigation on the bias and variance decomposition, the success of Canonical Forest can be attributed to the variance reduction.

Original languageEnglish
Pages (from-to)849-867
Number of pages19
JournalComputational Statistics
Volume29
Issue number3-4
DOIs
Publication statusPublished - 2014 Jan 1

Fingerprint

Discriminant analysis
Classifiers
Discriminant Analysis
Ensemble Methods
Classifier
Linear transformations
Variance Decomposition
Variance Reduction
Canonical Ensemble
Decomposition
Bootstrapping
Dimension Reduction
Linear transformation
Feature Space
Ensemble
Partial

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty
  • Computational Mathematics

Cite this

Chen, Y. C., Ha, H., Kim, H., & Ahn, H. (2014). Canonical Forest. Computational Statistics, 29(3-4), 849-867. https://doi.org/10.1007/s00180-013-0466-x
Chen, Yu Chuan ; Ha, Hyejung ; Kim, Hyunjoong ; Ahn, Hongshik. / Canonical Forest. In: Computational Statistics. 2014 ; Vol. 29, No. 3-4. pp. 849-867.
@article{dd0f811426d942aebcd81e979f006524,
title = "Canonical Forest",
abstract = "We propose a new classification ensemble method named Canonical Forest. The new method uses canonical linear discriminant analysis (CLDA) and bootstrapping to obtain accurate and diverse classifiers that constitute an ensemble. We note CLDA serves as a linear transformation tool rather than a dimension reduction tool. Since CLDA will find the transformed space that separates the classes farther in distribution, classifiers built on this space will be more accurate than those on the original space. To further facilitate the diversity of the classifiers in an ensemble, CLDA is applied only on a partial feature space for each bootstrapped data. To compare the performance of Canonical Forest and other widely used ensemble methods, we tested them on 29 real or artificial data sets. Canonical Forest performed significantly better in accuracy than other ensemble methods in most data sets. According to the investigation on the bias and variance decomposition, the success of Canonical Forest can be attributed to the variance reduction.",
author = "Chen, {Yu Chuan} and Hyejung Ha and Hyunjoong Kim and Hongshik Ahn",
year = "2014",
month = "1",
day = "1",
doi = "10.1007/s00180-013-0466-x",
language = "English",
volume = "29",
pages = "849--867",
journal = "Computational Statistics",
issn = "0943-4062",
publisher = "Springer Verlag",
number = "3-4",

}

Chen, YC, Ha, H, Kim, H & Ahn, H 2014, 'Canonical Forest', Computational Statistics, vol. 29, no. 3-4, pp. 849-867. https://doi.org/10.1007/s00180-013-0466-x

Canonical Forest. / Chen, Yu Chuan; Ha, Hyejung; Kim, Hyunjoong; Ahn, Hongshik.

In: Computational Statistics, Vol. 29, No. 3-4, 01.01.2014, p. 849-867.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Canonical Forest

AU - Chen, Yu Chuan

AU - Ha, Hyejung

AU - Kim, Hyunjoong

AU - Ahn, Hongshik

PY - 2014/1/1

Y1 - 2014/1/1

N2 - We propose a new classification ensemble method named Canonical Forest. The new method uses canonical linear discriminant analysis (CLDA) and bootstrapping to obtain accurate and diverse classifiers that constitute an ensemble. We note CLDA serves as a linear transformation tool rather than a dimension reduction tool. Since CLDA will find the transformed space that separates the classes farther in distribution, classifiers built on this space will be more accurate than those on the original space. To further facilitate the diversity of the classifiers in an ensemble, CLDA is applied only on a partial feature space for each bootstrapped data. To compare the performance of Canonical Forest and other widely used ensemble methods, we tested them on 29 real or artificial data sets. Canonical Forest performed significantly better in accuracy than other ensemble methods in most data sets. According to the investigation on the bias and variance decomposition, the success of Canonical Forest can be attributed to the variance reduction.

AB - We propose a new classification ensemble method named Canonical Forest. The new method uses canonical linear discriminant analysis (CLDA) and bootstrapping to obtain accurate and diverse classifiers that constitute an ensemble. We note CLDA serves as a linear transformation tool rather than a dimension reduction tool. Since CLDA will find the transformed space that separates the classes farther in distribution, classifiers built on this space will be more accurate than those on the original space. To further facilitate the diversity of the classifiers in an ensemble, CLDA is applied only on a partial feature space for each bootstrapped data. To compare the performance of Canonical Forest and other widely used ensemble methods, we tested them on 29 real or artificial data sets. Canonical Forest performed significantly better in accuracy than other ensemble methods in most data sets. According to the investigation on the bias and variance decomposition, the success of Canonical Forest can be attributed to the variance reduction.

UR - http://www.scopus.com/inward/record.url?scp=84901840344&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901840344&partnerID=8YFLogxK

U2 - 10.1007/s00180-013-0466-x

DO - 10.1007/s00180-013-0466-x

M3 - Article

AN - SCOPUS:84901840344

VL - 29

SP - 849

EP - 867

JO - Computational Statistics

JF - Computational Statistics

SN - 0943-4062

IS - 3-4

ER -