Table 4:
The results of Normalized mutual information (NMI±STD%) corresponding to five feature selection techniques computed on seven datasets. In every row, the first and the second best outcomes are boldfaced and are underscored, respectively. The number of the selected features for the best clustering outcomes is shown in parentheses. (A bigger value of NMI indicates a better clustering performance.)
| Dataset | Baseline | MFFS | MPMR | SGFS | RMFFS | SLSDR | |
|---|---|---|---|---|---|---|---|
| CNS | 1.17 ± 0.00 | 2.93 ± 0.45 (100) | 1.60 ± 0.80 (30) | 8.79 ± 0.00 (30) | 9.35 ± 0.08 (50) | 9.94 ± 3.66 (100) | |
| Colon Cancer | 27.35 ± 9.96 | 24.30 ± 0.14 (100) | 21.08 ± 0.54 (40) | 30.03 ± 0.00 (10) | 33.55 ± 0.04 (60) | 41.58 ± 0.03 (40) | |
| DLBCL | 2.93 ± 1.09 | 39.45 ± 0.00 (40) | 48.46 ± 0.72 (70) | 49.51 ± 0.72 (70) | 55.50 ± 0.65 (30) | 60.70 ± 1.24 (60) | |
| GLIOMA | 17.94 ± 0.69 | 23.69 ± 2.07 (80) | 25.24 ± 2.08 (30) | 27.59 ± 6.86 (90) | 29.81 ± 1.46 (30) | 32.42 ± 3.01 (100) | |
| Leukemia | 21.47 ± 0.00 | 66.56 ± 0.06 (80) | 62.07 ± 0.03 (30) | 72.87 ± 0.00 (30) | 80.05 ± 0.01 (70) | 83.78 ± 0.03 (30) | |
| Lung Cancer | 68.12 ± 0.56 | 55.61 ± 6.88 (100) | 55.77 ± 2.11 (70) | 65.71 ± 6.39 (100) | 59.71 ± 2.31 (70) | 69.48 ± 0.48 (70) | |
| Lymphoma | 69.55 ± 3.41 | 63.47 ± 3.03 (60) | 67.13 ± 3.12 (90) | 68.93 ± 3.12 (60) | 65.19 ± 2.29 (80) | 69.50 ± 1.90 (80) | |
| Prostate Tumor | 8.56 ± 0.21 | 6.09 ± 0.91 (60) | 8.83 ± 0.95 (20) | 9.45 ± 0.03 (80) | 11.50 ± 0.08 (40) | 12.83 ± 0.02 (30) | |
| SRBCT | 11.55 ± 3.47 | 38.41 ± 7.38 (50) | 24.27 ± 5.26 (40) | 48.37 ± 1.21 (10) | 47.82 ± 4.13 (90) | 50.33 ± 2.72 (50) | |
| TOX-171 | 13.54 ± 0.23 | 12.82 ± 1.52 (80) | 16.22 ± 1.89 (60) | 13.96 ± 1.02 (100) | 24.60 ± 0.41 (20) | 29.90 ± 0.69 (10) | |