Finding robust linear expression-based classifiers

S. Kim, E. R. Dougherty, J. Barrera, Yidong Chen, M. Bittner, J. M. Trent

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

A key goal for the use of gene-expression microarrays is to perform classification via different expression patterns. The typical small sample obtained and the large numbers of variables make the task of finding good classifiers extremely difficult, from the perspectives of both design and error estimation. This paper addresses the issue of estimation variability, which can result in large numbers of gene sets that have highly optimistic error estimates. It proposes performing classification on probability distributions derived from the original sample points by spreading the mass of those points to make classification more difficult while retaining the basic geometry of the point locations. This is done in a parameterized fashion, based on the degree to which the mass is spread. The method is applied to linear classifiers.

Original languageEnglish (US)
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
EditorsM.L. Bittner, Y. Chen, A.N. Dorsel, E.R. Dougherty
Pages207-212
Number of pages6
Volume4266
DOIs
StatePublished - 2001
Externally publishedYes
EventMicroarrays: Optical Technologies and Informatics - San Jose, CA, United States
Duration: Jan 21 2000Jan 22 2000

Other

OtherMicroarrays: Optical Technologies and Informatics
CountryUnited States
CitySan Jose, CA
Period1/21/001/22/00

Fingerprint

classifiers
Classifiers
gene expression
Microarrays
retaining
Gene expression
genes
Error analysis
Probability distributions
Genes
Geometry
estimates
geometry

Keywords

  • Classification
  • Genomics
  • Linear classifier
  • Microarray
  • Perceptron

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Condensed Matter Physics

Cite this

Kim, S., Dougherty, E. R., Barrera, J., Chen, Y., Bittner, M., & Trent, J. M. (2001). Finding robust linear expression-based classifiers. In M. L. Bittner, Y. Chen, A. N. Dorsel, & E. R. Dougherty (Eds.), Proceedings of SPIE - The International Society for Optical Engineering (Vol. 4266, pp. 207-212) https://doi.org/10.1117/12.427989

Finding robust linear expression-based classifiers. / Kim, S.; Dougherty, E. R.; Barrera, J.; Chen, Yidong; Bittner, M.; Trent, J. M.

Proceedings of SPIE - The International Society for Optical Engineering. ed. / M.L. Bittner; Y. Chen; A.N. Dorsel; E.R. Dougherty. Vol. 4266 2001. p. 207-212.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kim, S, Dougherty, ER, Barrera, J, Chen, Y, Bittner, M & Trent, JM 2001, Finding robust linear expression-based classifiers. in ML Bittner, Y Chen, AN Dorsel & ER Dougherty (eds), Proceedings of SPIE - The International Society for Optical Engineering. vol. 4266, pp. 207-212, Microarrays: Optical Technologies and Informatics, San Jose, CA, United States, 1/21/00. https://doi.org/10.1117/12.427989
Kim S, Dougherty ER, Barrera J, Chen Y, Bittner M, Trent JM. Finding robust linear expression-based classifiers. In Bittner ML, Chen Y, Dorsel AN, Dougherty ER, editors, Proceedings of SPIE - The International Society for Optical Engineering. Vol. 4266. 2001. p. 207-212 https://doi.org/10.1117/12.427989
Kim, S. ; Dougherty, E. R. ; Barrera, J. ; Chen, Yidong ; Bittner, M. ; Trent, J. M. / Finding robust linear expression-based classifiers. Proceedings of SPIE - The International Society for Optical Engineering. editor / M.L. Bittner ; Y. Chen ; A.N. Dorsel ; E.R. Dougherty. Vol. 4266 2001. pp. 207-212
@inproceedings{cbd99488f94442dc83cfa0bf1001c0f1,
title = "Finding robust linear expression-based classifiers",
abstract = "A key goal for the use of gene-expression microarrays is to perform classification via different expression patterns. The typical small sample obtained and the large numbers of variables make the task of finding good classifiers extremely difficult, from the perspectives of both design and error estimation. This paper addresses the issue of estimation variability, which can result in large numbers of gene sets that have highly optimistic error estimates. It proposes performing classification on probability distributions derived from the original sample points by spreading the mass of those points to make classification more difficult while retaining the basic geometry of the point locations. This is done in a parameterized fashion, based on the degree to which the mass is spread. The method is applied to linear classifiers.",
keywords = "Classification, Genomics, Linear classifier, Microarray, Perceptron",
author = "S. Kim and Dougherty, {E. R.} and J. Barrera and Yidong Chen and M. Bittner and Trent, {J. M.}",
year = "2001",
doi = "10.1117/12.427989",
language = "English (US)",
volume = "4266",
pages = "207--212",
editor = "M.L. Bittner and Y. Chen and A.N. Dorsel and E.R. Dougherty",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",

}

TY - GEN

T1 - Finding robust linear expression-based classifiers

AU - Kim, S.

AU - Dougherty, E. R.

AU - Barrera, J.

AU - Chen, Yidong

AU - Bittner, M.

AU - Trent, J. M.

PY - 2001

Y1 - 2001

N2 - A key goal for the use of gene-expression microarrays is to perform classification via different expression patterns. The typical small sample obtained and the large numbers of variables make the task of finding good classifiers extremely difficult, from the perspectives of both design and error estimation. This paper addresses the issue of estimation variability, which can result in large numbers of gene sets that have highly optimistic error estimates. It proposes performing classification on probability distributions derived from the original sample points by spreading the mass of those points to make classification more difficult while retaining the basic geometry of the point locations. This is done in a parameterized fashion, based on the degree to which the mass is spread. The method is applied to linear classifiers.

AB - A key goal for the use of gene-expression microarrays is to perform classification via different expression patterns. The typical small sample obtained and the large numbers of variables make the task of finding good classifiers extremely difficult, from the perspectives of both design and error estimation. This paper addresses the issue of estimation variability, which can result in large numbers of gene sets that have highly optimistic error estimates. It proposes performing classification on probability distributions derived from the original sample points by spreading the mass of those points to make classification more difficult while retaining the basic geometry of the point locations. This is done in a parameterized fashion, based on the degree to which the mass is spread. The method is applied to linear classifiers.

KW - Classification

KW - Genomics

KW - Linear classifier

KW - Microarray

KW - Perceptron

UR - http://www.scopus.com/inward/record.url?scp=0034866682&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034866682&partnerID=8YFLogxK

U2 - 10.1117/12.427989

DO - 10.1117/12.427989

M3 - Conference contribution

AN - SCOPUS:0034866682

VL - 4266

SP - 207

EP - 212

BT - Proceedings of SPIE - The International Society for Optical Engineering

A2 - Bittner, M.L.

A2 - Chen, Y.

A2 - Dorsel, A.N.

A2 - Dougherty, E.R.

ER -