Improved classification accuracy by feature extraction using genetic algorithms

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

1 Citation (Scopus)

Abstract

A feature extraction algorithm has been developed for the purposes of improving classification accuracy. The algorithm uses a genetic algorithm / hill-climber hybrid to generate a set of linearly recombined features, which may be of reduced dimensionality compared with the original set. The genetic algorithm performs the global exploration, and a hill climber explores local neighborhoods. Hybridizing the genetic algorithm with a hill climber improves both the rate of convergence, and the final overall cost function value; it also reduces the sensitivity of the genetic algorithm to parameter selection. The genetic algorithm includes the operators: crossover, mutation, and deletion / reactivation - the last of these effects dimensionality reduction. The feature extractor is supervised, and is capable of deriving a separate feature space for each tissue (which are reintegrated during classification). A non-anatomical digital phantom was developed as a gold standard for testing purposes. In tests with the phantom, and with images of multiple sclerosis patients, classification with feature extractor derived features yielded lower error rates than using standard pulse sequences, and with features derived using principal components analysis. Using the multiple sclerosis patient data, the algorithm resulted in a mean 31% reduction in classification error of pure tissues.

Original languageEnglish (US)
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
EditorsM. Sonka, J.M. Fitzpatrick
Pages1402-1412
Number of pages11
Volume5032 III
DOIs
StatePublished - 2003
EventMedical Imaging 2003: Image Processing - San Diego, CA, United States
Duration: Feb 17 2003Feb 20 2003

Other

OtherMedical Imaging 2003: Image Processing
CountryUnited States
CitySan Diego, CA
Period2/17/032/20/03

Fingerprint

genetic algorithms
pattern recognition
Feature extraction
Genetic algorithms
Tissue
deletion
mutations
principal components analysis
Cost functions
Principal component analysis
Mathematical operators
crossovers
costs
operators
sensitivity
Testing
pulses

Keywords

  • Classification
  • Feature extraction
  • Genetic algorithm

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Condensed Matter Physics

Cite this

Patriarche, J., Manduca, A., & Erickson, B. J. (2003). Improved classification accuracy by feature extraction using genetic algorithms. In M. Sonka, & J. M. Fitzpatrick (Eds.), Proceedings of SPIE - The International Society for Optical Engineering (Vol. 5032 III, pp. 1402-1412) https://doi.org/10.1117/12.481397

Improved classification accuracy by feature extraction using genetic algorithms. / Patriarche, Julia; Manduca, Armando; Erickson, Bradley J.

Proceedings of SPIE - The International Society for Optical Engineering. ed. / M. Sonka; J.M. Fitzpatrick. Vol. 5032 III 2003. p. 1402-1412.

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

Patriarche, J, Manduca, A & Erickson, BJ 2003, Improved classification accuracy by feature extraction using genetic algorithms. in M Sonka & JM Fitzpatrick (eds), Proceedings of SPIE - The International Society for Optical Engineering. vol. 5032 III, pp. 1402-1412, Medical Imaging 2003: Image Processing, San Diego, CA, United States, 2/17/03. https://doi.org/10.1117/12.481397
Patriarche J, Manduca A, Erickson BJ. Improved classification accuracy by feature extraction using genetic algorithms. In Sonka M, Fitzpatrick JM, editors, Proceedings of SPIE - The International Society for Optical Engineering. Vol. 5032 III. 2003. p. 1402-1412 https://doi.org/10.1117/12.481397
Patriarche, Julia ; Manduca, Armando ; Erickson, Bradley J. / Improved classification accuracy by feature extraction using genetic algorithms. Proceedings of SPIE - The International Society for Optical Engineering. editor / M. Sonka ; J.M. Fitzpatrick. Vol. 5032 III 2003. pp. 1402-1412
@inproceedings{021477bc6b2d41f78c728b1acdce19f1,
title = "Improved classification accuracy by feature extraction using genetic algorithms",
abstract = "A feature extraction algorithm has been developed for the purposes of improving classification accuracy. The algorithm uses a genetic algorithm / hill-climber hybrid to generate a set of linearly recombined features, which may be of reduced dimensionality compared with the original set. The genetic algorithm performs the global exploration, and a hill climber explores local neighborhoods. Hybridizing the genetic algorithm with a hill climber improves both the rate of convergence, and the final overall cost function value; it also reduces the sensitivity of the genetic algorithm to parameter selection. The genetic algorithm includes the operators: crossover, mutation, and deletion / reactivation - the last of these effects dimensionality reduction. The feature extractor is supervised, and is capable of deriving a separate feature space for each tissue (which are reintegrated during classification). A non-anatomical digital phantom was developed as a gold standard for testing purposes. In tests with the phantom, and with images of multiple sclerosis patients, classification with feature extractor derived features yielded lower error rates than using standard pulse sequences, and with features derived using principal components analysis. Using the multiple sclerosis patient data, the algorithm resulted in a mean 31{\%} reduction in classification error of pure tissues.",
keywords = "Classification, Feature extraction, Genetic algorithm",
author = "Julia Patriarche and Armando Manduca and Erickson, {Bradley J}",
year = "2003",
doi = "10.1117/12.481397",
language = "English (US)",
volume = "5032 III",
pages = "1402--1412",
editor = "M. Sonka and J.M. Fitzpatrick",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",

}

TY - GEN

T1 - Improved classification accuracy by feature extraction using genetic algorithms

AU - Patriarche, Julia

AU - Manduca, Armando

AU - Erickson, Bradley J

PY - 2003

Y1 - 2003

N2 - A feature extraction algorithm has been developed for the purposes of improving classification accuracy. The algorithm uses a genetic algorithm / hill-climber hybrid to generate a set of linearly recombined features, which may be of reduced dimensionality compared with the original set. The genetic algorithm performs the global exploration, and a hill climber explores local neighborhoods. Hybridizing the genetic algorithm with a hill climber improves both the rate of convergence, and the final overall cost function value; it also reduces the sensitivity of the genetic algorithm to parameter selection. The genetic algorithm includes the operators: crossover, mutation, and deletion / reactivation - the last of these effects dimensionality reduction. The feature extractor is supervised, and is capable of deriving a separate feature space for each tissue (which are reintegrated during classification). A non-anatomical digital phantom was developed as a gold standard for testing purposes. In tests with the phantom, and with images of multiple sclerosis patients, classification with feature extractor derived features yielded lower error rates than using standard pulse sequences, and with features derived using principal components analysis. Using the multiple sclerosis patient data, the algorithm resulted in a mean 31% reduction in classification error of pure tissues.

AB - A feature extraction algorithm has been developed for the purposes of improving classification accuracy. The algorithm uses a genetic algorithm / hill-climber hybrid to generate a set of linearly recombined features, which may be of reduced dimensionality compared with the original set. The genetic algorithm performs the global exploration, and a hill climber explores local neighborhoods. Hybridizing the genetic algorithm with a hill climber improves both the rate of convergence, and the final overall cost function value; it also reduces the sensitivity of the genetic algorithm to parameter selection. The genetic algorithm includes the operators: crossover, mutation, and deletion / reactivation - the last of these effects dimensionality reduction. The feature extractor is supervised, and is capable of deriving a separate feature space for each tissue (which are reintegrated during classification). A non-anatomical digital phantom was developed as a gold standard for testing purposes. In tests with the phantom, and with images of multiple sclerosis patients, classification with feature extractor derived features yielded lower error rates than using standard pulse sequences, and with features derived using principal components analysis. Using the multiple sclerosis patient data, the algorithm resulted in a mean 31% reduction in classification error of pure tissues.

KW - Classification

KW - Feature extraction

KW - Genetic algorithm

UR - http://www.scopus.com/inward/record.url?scp=0042421857&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0042421857&partnerID=8YFLogxK

U2 - 10.1117/12.481397

DO - 10.1117/12.481397

M3 - Conference contribution

VL - 5032 III

SP - 1402

EP - 1412

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

A2 - Sonka, M.

A2 - Fitzpatrick, J.M.

ER -