A two-population evolutionary algorithm for feature extraction: Combining filter and wrapper

Eun Yeong Ahn, Tracy Mullen, John Yen

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

Abstract

Extracting good features is critical to the performance of learning algorithms such as classifiers. Feature extraction selects and transforms original features to find information hidden in data. Due to the huge search space of selection and transformation of features, exhaustive search is computationally prohibitive and randomized search such as evolutionary algorithms (EA) are often used. In our prior work on evolutionary-based feature extraction, an individual, which represents a set of features, is evaluated by estimating the accuracy of a classifier when the individual's feature set is used for learning. Although incorporating a learning algorithm during evaluation, which is called the wrapper approach, generally performs better than evaluating an individual simply by the statistical properties of data, which is called the filter appproach, our EA based on a wrapper approach suffers from overfitting, so that a slight enhancement of fitness in training can dramatically reduce the classification accuracy for unseen testing data. To cope with this problem, this paper proposes a two-population EA for feature extraction (TEAFE) that combines filter and wrapper approaches, and shows the promising preliminary results.

Original languageEnglish (US)
Title of host publication2011 IEEE Congress of Evolutionary Computation, CEC 2011
Pages736-743
Number of pages8
DOIs
StatePublished - 2011
Event2011 IEEE Congress of Evolutionary Computation, CEC 2011 - New Orleans, LA, United States
Duration: Jun 5 2011Jun 8 2011

Other

Other2011 IEEE Congress of Evolutionary Computation, CEC 2011
CountryUnited States
CityNew Orleans, LA
Period6/5/116/8/11

Fingerprint

Wrapper
Evolutionary algorithms
Feature Extraction
Evolutionary Algorithms
Feature extraction
Filter
Learning algorithms
Learning Algorithm
Classifiers
Classifier
Overfitting
Exhaustive Search
Statistical property
Search Space
Fitness
Enhancement
Transform
Testing
Evaluation

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Theoretical Computer Science

Cite this

Ahn, E. Y., Mullen, T., & Yen, J. (2011). A two-population evolutionary algorithm for feature extraction: Combining filter and wrapper. In 2011 IEEE Congress of Evolutionary Computation, CEC 2011 (pp. 736-743). [5949692] https://doi.org/10.1109/CEC.2011.5949692
Ahn, Eun Yeong ; Mullen, Tracy ; Yen, John. / A two-population evolutionary algorithm for feature extraction : Combining filter and wrapper. 2011 IEEE Congress of Evolutionary Computation, CEC 2011. 2011. pp. 736-743
@inproceedings{074743b5939f4247b138196f6c058198,
title = "A two-population evolutionary algorithm for feature extraction: Combining filter and wrapper",
abstract = "Extracting good features is critical to the performance of learning algorithms such as classifiers. Feature extraction selects and transforms original features to find information hidden in data. Due to the huge search space of selection and transformation of features, exhaustive search is computationally prohibitive and randomized search such as evolutionary algorithms (EA) are often used. In our prior work on evolutionary-based feature extraction, an individual, which represents a set of features, is evaluated by estimating the accuracy of a classifier when the individual's feature set is used for learning. Although incorporating a learning algorithm during evaluation, which is called the wrapper approach, generally performs better than evaluating an individual simply by the statistical properties of data, which is called the filter appproach, our EA based on a wrapper approach suffers from overfitting, so that a slight enhancement of fitness in training can dramatically reduce the classification accuracy for unseen testing data. To cope with this problem, this paper proposes a two-population EA for feature extraction (TEAFE) that combines filter and wrapper approaches, and shows the promising preliminary results.",
author = "Ahn, {Eun Yeong} and Tracy Mullen and John Yen",
year = "2011",
doi = "10.1109/CEC.2011.5949692",
language = "English (US)",
isbn = "9781424478347",
pages = "736--743",
booktitle = "2011 IEEE Congress of Evolutionary Computation, CEC 2011",

}

Ahn, EY, Mullen, T & Yen, J 2011, A two-population evolutionary algorithm for feature extraction: Combining filter and wrapper. in 2011 IEEE Congress of Evolutionary Computation, CEC 2011., 5949692, pp. 736-743, 2011 IEEE Congress of Evolutionary Computation, CEC 2011, New Orleans, LA, United States, 6/5/11. https://doi.org/10.1109/CEC.2011.5949692

A two-population evolutionary algorithm for feature extraction : Combining filter and wrapper. / Ahn, Eun Yeong; Mullen, Tracy; Yen, John.

2011 IEEE Congress of Evolutionary Computation, CEC 2011. 2011. p. 736-743 5949692.

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

TY - GEN

T1 - A two-population evolutionary algorithm for feature extraction

T2 - Combining filter and wrapper

AU - Ahn, Eun Yeong

AU - Mullen, Tracy

AU - Yen, John

PY - 2011

Y1 - 2011

N2 - Extracting good features is critical to the performance of learning algorithms such as classifiers. Feature extraction selects and transforms original features to find information hidden in data. Due to the huge search space of selection and transformation of features, exhaustive search is computationally prohibitive and randomized search such as evolutionary algorithms (EA) are often used. In our prior work on evolutionary-based feature extraction, an individual, which represents a set of features, is evaluated by estimating the accuracy of a classifier when the individual's feature set is used for learning. Although incorporating a learning algorithm during evaluation, which is called the wrapper approach, generally performs better than evaluating an individual simply by the statistical properties of data, which is called the filter appproach, our EA based on a wrapper approach suffers from overfitting, so that a slight enhancement of fitness in training can dramatically reduce the classification accuracy for unseen testing data. To cope with this problem, this paper proposes a two-population EA for feature extraction (TEAFE) that combines filter and wrapper approaches, and shows the promising preliminary results.

AB - Extracting good features is critical to the performance of learning algorithms such as classifiers. Feature extraction selects and transforms original features to find information hidden in data. Due to the huge search space of selection and transformation of features, exhaustive search is computationally prohibitive and randomized search such as evolutionary algorithms (EA) are often used. In our prior work on evolutionary-based feature extraction, an individual, which represents a set of features, is evaluated by estimating the accuracy of a classifier when the individual's feature set is used for learning. Although incorporating a learning algorithm during evaluation, which is called the wrapper approach, generally performs better than evaluating an individual simply by the statistical properties of data, which is called the filter appproach, our EA based on a wrapper approach suffers from overfitting, so that a slight enhancement of fitness in training can dramatically reduce the classification accuracy for unseen testing data. To cope with this problem, this paper proposes a two-population EA for feature extraction (TEAFE) that combines filter and wrapper approaches, and shows the promising preliminary results.

UR - http://www.scopus.com/inward/record.url?scp=80052002589&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052002589&partnerID=8YFLogxK

U2 - 10.1109/CEC.2011.5949692

DO - 10.1109/CEC.2011.5949692

M3 - Conference contribution

AN - SCOPUS:80052002589

SN - 9781424478347

SP - 736

EP - 743

BT - 2011 IEEE Congress of Evolutionary Computation, CEC 2011

ER -

Ahn EY, Mullen T, Yen J. A two-population evolutionary algorithm for feature extraction: Combining filter and wrapper. In 2011 IEEE Congress of Evolutionary Computation, CEC 2011. 2011. p. 736-743. 5949692 https://doi.org/10.1109/CEC.2011.5949692