Skip to main content

Table 3 Performance comparison for multi-class datasets.

From: Recursive gene selection based on maximum margin criterion: a comparison with SVM-RFE

Classifier+Selection criterion

Number of genes

 

10

20

30

50

100

MLL

     

NMC+BW

11.5 ± 0.7

8.8 ± 0.6

7.4 ± 0.5

6.1 ± 0.5

5.6 ± 0.5

NMC+MMC-RFE(U)

7.0 ± 0.6

5.8 ± 0.5

5.1 ± 0.5

4.9 ± 0.5

4.0 ± 0.4

NMC+MMC-RFE(O)

6.4 ± 0.5

5.9 ± 0.5

5.6 ± 0.5

4.9 ± 0.4

4.4 ± 0.4

NMC+SVM-RFE(H)

26.9 ± 1.4

19.3 ± 1.2

15.5 ± 1.1

12.0 ± 0.8

9.1 ± 0.7

NMC+SVM-RFE(S)

28.0 ± 1.3

21.4 ± 1.1

16.6 ± 1.0

11.9 ± 0.8

7.9 ± 0.7

MMC+MMC-RFE(U)

6.8 ± 0.5

6.0 ± 0.5

5.2 ± 0.5

4.9 ± 0.5

4.0 ± 0.4

MMC+MMC-RFE(O)

6.4 ± 0.5

5.8 ± 0.5

5.6 ± 0.5

4.9 ± 0.4

4.5 ± 0.4

SVM+SVM-RFE(H)

31.3 ± 1.5

24.0 ± 1.4

18.3 ± 1.1

12.9 ± 0.8

7.9 ± 0.6

SVM+SVM-RFE(S)

26.2 ± 1.2

20.2 ± 1.1

14.4 ± 1.0

10.6 ± 0.8

6.8 ± 0.6

SRBCT

     

NMC+BW

35.2 ± 1.4

22.1 ± 0.7

19.3 ± 0.7

10.5 ± 0.7

7.6 ± 0.6

NMC+MMC-RFE(U)

5.0 ± 0.5

3.0 ± 0.4

2.4 ± 0.3

2.2 ± 0.3

2.7 ± 0.3

NMC+MMC-RFE(O)

8.9 ± 0.7

6.0 ± 0.5

6.5 ± 0.5

6.8 ± 0.5

6.4 ± 0.5

NMC+SVM-RFE(H)

29.2 ± 1.2

22.9 ± 1.1

19.5 ± 1.0

15.7 ± 0.9

11.6 ± 0.7

NMC+SVM-RFE(S)

27.2 ± 1.2

21.9 ± 1.2

18.3 ± 1.0

14.2 ± 0.7

11.1 ± 0.8

MMC+MMC-RFE(U)

4.4 ± 0.5

2.5 ± 0.3

2.0 ± 0.3

1.7 ± 0.3

1.3 ± 0.2

MMC+MMC-RFE(O)

4.7 ± 0.5

4.1 ± 0.4

4.4 ± 0.4

3.5 ± 0.4

3.3 ± 0.4

SVM+SVM-RFE(H)

24.0 ± 1.3

14.2 ± 1.0

9.6 ± 0.7

6.3 ± 0.5

3.6 ± 0.4

SVM+SVM-RFE(S)

24.8 ± 1.4

12.7 ± 1.1

8.8 ± 0.8

5.1 ± 0.5

3.4 ± 0.4

  1. The average error and standard error rates (%) for MLL and SRBCT, when the number of genes is {10, 20, 30, 50, 100}. SVM-RFE(S) shows the best result with respect to the C parameter; NMC+SVM-RFE(S): C = 0.1, SVM+SVM-RFE(S): C = 0.1 for MLL; NMC+SVM-RFE(S): C = 100, SVM+SVM-RFE(S): C = 1000 for SRBCT.