Int. J. Fuzzy Log. Intell. Syst. 2018; 18(2): 120-125
Published online June 25, 2018
https://doi.org/10.5391/IJFIS.2018.18.2.120
© The Korean Institute of Intelligent Systems
Heesung Lee
Department of Railroad Electrical and Electronics Engineering, Korea National University of Transportation, Uiwang-si, Korea
Correspondence to :
Heesung Lee (hslee0717@ut.ac.kr)
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper proposes a new feature extraction scheme, combining global and local features. The proposed method uses principal component analysis (PCA) and linear discriminant analysis (LDA) for global property and locality preserving projections (LPP) for local property of the pattern. PCA and LDA are known for extracting the most descriptive ones after projection while LPP is known for preserving the neighborhood structure of the data set. The proposed combing method integrates global and local descriptive information and finds an efficient set of alternatives beyond PCA, LDA and LPP in the parametric space. Further, In order to find optimal parameters, the genetic algorithm (GA) is employed. Experiments are performed with four data sets selected in UCI machine learning repository to show the performance of the proposed algorithm.
Keywords: PCA, LDA, LPP, GA, Feature space, UCI machine learning repository
No potential conflict of Interest relevant to this article was reported.
E-mail: hslee0717@ut.ac.kr
Int. J. Fuzzy Log. Intell. Syst. 2018; 18(2): 120-125
Published online June 25, 2018 https://doi.org/10.5391/IJFIS.2018.18.2.120
Copyright © The Korean Institute of Intelligent Systems.
Heesung Lee
Department of Railroad Electrical and Electronics Engineering, Korea National University of Transportation, Uiwang-si, Korea
Correspondence to:Heesung Lee (hslee0717@ut.ac.kr)
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper proposes a new feature extraction scheme, combining global and local features. The proposed method uses principal component analysis (PCA) and linear discriminant analysis (LDA) for global property and locality preserving projections (LPP) for local property of the pattern. PCA and LDA are known for extracting the most descriptive ones after projection while LPP is known for preserving the neighborhood structure of the data set. The proposed combing method integrates global and local descriptive information and finds an efficient set of alternatives beyond PCA, LDA and LPP in the parametric space. Further, In order to find optimal parameters, the genetic algorithm (GA) is employed. Experiments are performed with four data sets selected in UCI machine learning repository to show the performance of the proposed algorithm.
Keywords: PCA, LDA, LPP, GA, Feature space, UCI machine learning repository
Structure of the chromosome.
Correct classification ratio for four databases of the UCI repository.
procedure Genetic Algorithm |
begin |
initialize |
while termination-condition not satisfied do |
begin |
evaluate |
select |
crossover |
mutation |
|
end |
end |
Table 1. Databases used for experimnets.
Database | Number of instances | Number of classes | Number of features |
---|---|---|---|
Sonar | 208 | 3 | 60 |
Spambase | 4601 | 2 | 57 |
SPECTF | 267 | 2 | 44 |
Heart | 270 | 2 | 13 |
Table 2. GA parameters.
Parameter | Value |
---|---|
Crossover rate | 0.6 |
Mutation rate | 0.05 |
Population size | 100 |
Generation | 200 |
Table 3. Average, the best, and the worst accrucy using PCA.
Database | Average | Best | Worst |
---|---|---|---|
Sonar | 82.35 (0.053) | 86.27 | 74.51 |
Spambase | 81.84 (0.004) | 82.35 | 81.39 |
SPECTF | 69.70 (0.081) | 74.24 | 57.58 |
Heart | 62.31 (0.022) | 64.17 | 59.70 |
Table 4. Average, the best, and the worst accrucy using LDA.
Database | Average | Best | Worst |
---|---|---|---|
Sonar | 74.02 (0.079) | 80.40 | 62.75 |
Spambase | 78.37 (0.022) | 81.65 | 77.13 |
SPECTF | 73.11 (0.054) | 75.76 | 65.15 |
Heart | 53.36 (0.056) | 59.70 | 46.27 |
Table 5. Average, the best, and the worst accrucy using LPP.
Database | Average | Best | Worst |
---|---|---|---|
Sonar | 84.80 (0.040) | 90.20 | 80.39 |
Spambase | 82.04 (0.002) | 82.26 | 81.82 |
SPECTF | 73.86 (0.050) | 80.30 | 68.19 |
Heart | 55.97 (0.019) | 58.21 | 53.73 |
Table 6. Average, the best, and the worst accrucy of the proposed method.
Database | Average | Best | Worst |
---|---|---|---|
Sonar | 86.27 (0.679) | 92.16 | 76.47 |
Spambase | 83.13 (0.009) | 83.57 | 82.26 |
SPECTF | 75.37 (0.084) | 83.33 | 63.64 |
Heart | 63.43 (0.026) | 65.67 | 59.70 |
Erdenebayar Urtnasan, Jong-Uk Park, SooYong Lee, and Kyoung-Joung Lee
Int. J. Fuzzy Log. Intell. Syst. -0001; 17(2): 76-81 https://doi.org/10.5391/IJFIS.2017.17.2.76Lae-Jeong Park
Int. J. Fuzzy Log. Intell. Syst. 2010; 10(2): 95-100Sung-Ho Kim,Young-Sam Lee
Int. J. Fuzzy Log. Intell. Syst. 2006; 6(3): 217-222Structure of the chromosome.
|@|~(^,^)~|@|Correct classification ratio for four databases of the UCI repository.