A genetic feature weighting scheme for pattern recognition

Heesung Lee, Euntai Kim, Mignon Park

Research output: Contribution to journalArticle

32 Citations (Scopus)

Abstract

This paper proposes a new pattern recognition scheme, combining a new adaptive feature weighting and modified k-Nearest Neighbor (k-NN) rule. The proposed feature weighting method named adaptive-3FW. It uses three non-uniform weight levels (zero weight, middle weight and full weight) to weight each feature. The middle weight value is determined using genetic algorithms (GAs). The proposed adaptive-3FW overcomes overfitting issues and achieves high recognition performance. Novel GA operators tailored for this formulation are introduced to implement the proposed scheme. Further, a modified k-NN is proposed which uses a class-dependent feature weighting strategy. Whilst the conventional pattern recognition systems use the same set of feature weights for all classes, the proposed algorithm uses different sets of feature weights for different classes. Experiments were performed with the UCI repository for machine learning databases and the unconstrained handwritten numeral database of Concordia University in Canada to show the performance of the proposed method.

Original languageEnglish
Pages (from-to)161-171
Number of pages11
JournalIntegrated Computer-Aided Engineering
Volume14
Issue number2
Publication statusPublished - 2007 Aug 7

Fingerprint

Feature Weighting
Pattern Recognition
Pattern recognition
Genetic algorithms
Pattern recognition systems
Learning systems
Mathematical operators
Nearest Neighbor
Experiments
Genetic Algorithm
Numeral
Overfitting
Repository
Machine Learning
Formulation
Dependent

All Science Journal Classification (ASJC) codes

  • Software
  • Theoretical Computer Science
  • Computer Science Applications
  • Computational Theory and Mathematics
  • Artificial Intelligence

Cite this

Lee, Heesung ; Kim, Euntai ; Park, Mignon. / A genetic feature weighting scheme for pattern recognition. In: Integrated Computer-Aided Engineering. 2007 ; Vol. 14, No. 2. pp. 161-171.
@article{b9b82a2353fd4f45a3de183d043ca377,
title = "A genetic feature weighting scheme for pattern recognition",
abstract = "This paper proposes a new pattern recognition scheme, combining a new adaptive feature weighting and modified k-Nearest Neighbor (k-NN) rule. The proposed feature weighting method named adaptive-3FW. It uses three non-uniform weight levels (zero weight, middle weight and full weight) to weight each feature. The middle weight value is determined using genetic algorithms (GAs). The proposed adaptive-3FW overcomes overfitting issues and achieves high recognition performance. Novel GA operators tailored for this formulation are introduced to implement the proposed scheme. Further, a modified k-NN is proposed which uses a class-dependent feature weighting strategy. Whilst the conventional pattern recognition systems use the same set of feature weights for all classes, the proposed algorithm uses different sets of feature weights for different classes. Experiments were performed with the UCI repository for machine learning databases and the unconstrained handwritten numeral database of Concordia University in Canada to show the performance of the proposed method.",
author = "Heesung Lee and Euntai Kim and Mignon Park",
year = "2007",
month = "8",
day = "7",
language = "English",
volume = "14",
pages = "161--171",
journal = "Integrated Computer-Aided Engineering",
issn = "1069-2509",
publisher = "IOS Press",
number = "2",

}

A genetic feature weighting scheme for pattern recognition. / Lee, Heesung; Kim, Euntai; Park, Mignon.

In: Integrated Computer-Aided Engineering, Vol. 14, No. 2, 07.08.2007, p. 161-171.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A genetic feature weighting scheme for pattern recognition

AU - Lee, Heesung

AU - Kim, Euntai

AU - Park, Mignon

PY - 2007/8/7

Y1 - 2007/8/7

N2 - This paper proposes a new pattern recognition scheme, combining a new adaptive feature weighting and modified k-Nearest Neighbor (k-NN) rule. The proposed feature weighting method named adaptive-3FW. It uses three non-uniform weight levels (zero weight, middle weight and full weight) to weight each feature. The middle weight value is determined using genetic algorithms (GAs). The proposed adaptive-3FW overcomes overfitting issues and achieves high recognition performance. Novel GA operators tailored for this formulation are introduced to implement the proposed scheme. Further, a modified k-NN is proposed which uses a class-dependent feature weighting strategy. Whilst the conventional pattern recognition systems use the same set of feature weights for all classes, the proposed algorithm uses different sets of feature weights for different classes. Experiments were performed with the UCI repository for machine learning databases and the unconstrained handwritten numeral database of Concordia University in Canada to show the performance of the proposed method.

AB - This paper proposes a new pattern recognition scheme, combining a new adaptive feature weighting and modified k-Nearest Neighbor (k-NN) rule. The proposed feature weighting method named adaptive-3FW. It uses three non-uniform weight levels (zero weight, middle weight and full weight) to weight each feature. The middle weight value is determined using genetic algorithms (GAs). The proposed adaptive-3FW overcomes overfitting issues and achieves high recognition performance. Novel GA operators tailored for this formulation are introduced to implement the proposed scheme. Further, a modified k-NN is proposed which uses a class-dependent feature weighting strategy. Whilst the conventional pattern recognition systems use the same set of feature weights for all classes, the proposed algorithm uses different sets of feature weights for different classes. Experiments were performed with the UCI repository for machine learning databases and the unconstrained handwritten numeral database of Concordia University in Canada to show the performance of the proposed method.

UR - http://www.scopus.com/inward/record.url?scp=34547558060&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34547558060&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:34547558060

VL - 14

SP - 161

EP - 171

JO - Integrated Computer-Aided Engineering

JF - Integrated Computer-Aided Engineering

SN - 1069-2509

IS - 2

ER -