Table 3. SVM Precision(P), Recall(R) and F-measure(F) on top 20 features using 5-fold CV on different data sets.
TFS methods are compared against our methods (SMBA and SMBA-CSFS). FS: Fisher Score, mRMR: Minimum-Redundancy-Maximum-Relevance, MI: Mutual Information, RFS: Robust Feature Selector, EN: Elastic Net, BSL: all features. The best results are highlighted in bold. The number in parentheses is the number of features when the performance is achieved.
| ALLAML | LEUKEMIA | CLL_SUB_111 | GLIOMA | LUNG_C | LUNG_D | DLBCL | CARCINOM | GCM(14) | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| P | R | F | P | R | F | P | R | F | P | R | F | P | R | F | P | R | F | P | R | F | P | R | F | P | R | F | |
| Fisher | 0.98(18) | 0.98(18) | 0.98 | 0.99(15) | 0.99(15) | 0.99 | 0.75(11) | 0.75(11) | 0.75 | 0.68(20) | 0.67(14) | 0.67 | 0.92(19) | 0.92(19) | 0.92 | 0.89(20) | 0.88(15) | 0.88 | 0.9(17) | 0.99(20) | 0.93 | 0.9(19) | 0.89(20) | 0.89 | 0.64(20) | 0.64(20) | 0.64 |
| Relief | 0.96(12) | 0.96(12) | 0.96 | 0.99(4) | 0.99(4) | 0.99 | 0.75(17) | 0.75(17) | 0.75 | 0.77(19) | 0.77(19) | 0.77 | 0.97(20) | 0.97(20) | 0.97 | 0.95(20) | 0.95(15) | 0.95 | 0.89(18) | 1.0(20) | 0.94 | 0.89(18) | 0.88(18) | 0.88 | 0.8(20) | 0.8(20) | 0.8 |
| mRMR | 0.8(19) | 0.8(19) | 0.8 | 0.98(6) | 0.98(17) | 0.98 | 0.64(14) | 0.66(14) | 0.65 | 0.7(12) | 0.7(12) | 0.7 | 0.97(20) | 0.97(20) | 0.97 | 0.96(19) | 0.95(19) | 0.95 | 0.95(20) | 0.99(14) | 0.92 | 0.88(20) | 0.91(20) | 0.89 | 0.85(20) | 0.85(20) | 0.85 |
| MI | 0.98(12) | 0.98(12) | 0.98 | 0.98(2) | 0.98(2) | 0.98 | 0.76(16) | 0.76(16) | 0.76 | 0.74(20) | 0.73(17) | 0.73 | 0.97(20) | 0.97(20) | 0.97 | 0.95(20) | 0.95(20) | 0.95 | 0.95(17) | 0.99(19) | 0.9 | 0.95(17) | 0.95(17) | 0.83 | 0.69(20) | 0.69(20) | 0.69 |
| ls_l21 | 0.83(18) | 0.81(18) | 0.82 | 0.84(20) | 0.82(20) | 0.83 | 0.7(20) | 0.7(20) | 0.7 | 0.7(16) | 0.7(17) | 0.7 | 0.97(20) | 0.97(20) | 0.97 | 0.89(19) | 0.88(19) | 0.88 | 0.81(19) | 0.93(17) | 0.87 | 0.81(19) | 0.81(20) | 0.81 | 0.76(20) | 0.76(20) | 0.76 |
| ll_l21 | 0.92(15) | 0.91(15) | 0.91 | 0.83(20) | 0.83(20) | 0.83 | 0.69(20) | 0.69(20) | 0.69 | 0.65(9) | 0.65(9) | 0.65 | 0.98(18) | 0.98(18) | 0.98 | 0.94(20) | 0.93(20) | 0.93 | 0.92(18) | 0.96(19) | 0.92 | 0.9(17) | 0.86(20) | 0.88 | 0.84(20) | 0.84(20) | 0.84 |
| RFS | 0.86(18) | 0.84(19) | 0.85 | 0.84(20) | 0.76(20) | 0.8 | 0.63(12) | 0.64(12) | 0.63 | 0.71(12) | 0.7(12) | 0.7 | 0.96(19) | 0.96(19) | 0.96 | 0.88(18) | 0.86(18) | 0.87 | 0.89(19) | 0.93(16) | 0.84 | 0.89(18) | 0.84(19) | 0.86 | 0.77(20) | 0.77(20) | 0.77 |
| LASSO | 0.84(20) | 0.84(13) | 0.84 | 0.77(20) | 0.77(20) | 0.77 | 0.71(6) | 0.71(10) | 0.71 | 0.79(14) | 0.78(14) | 0.78 | 0.94(20) | 0.94(19) | 0.94 | 0.93(19) | 0.9(20) | 0.91 | 0.84(18) | 0.97(19) | 0.9 | 0.84(18) | 0.84(18) | 0.84 | 0.8(20) | 0.8(20) | 0.8 |
| EN | 0.84(20) | 0.84(13) | 0.84 | 0.77(20) | 0.77(20) | 0.77 | 0.71(6) | 0.71(10) | 0.71 | 0.79(14) | 0.78(14) | 0.78 | 0.94(20) | 0.94(19) | 0.94 | 0.91(19) | 0.9(20) | 0.9 | 0.84(18) | 0.97(19) | 0.9 | 0.84(18) | 0.84(18) | 0.84 | 0.8(20) | 0.8(20) | 0.8 |
| SMBA | 0.9(13) | 0.89(16) | 0.89 | 0.83(20) | 0.83(20) | 0.83 | 0.7(11) | 0.7(11) | 0.7 | 0.68(15) | 0.68(15) | 0.68 | 0.97(18) | 0.97(18) | 0.97 | 0.91(19) | 0.9(19) | 0.9 | 0.92(19) | 0.99(17) | 0.92 | 0.9(19) | 0.86(20) | 0.88 | 0.84(20) | 0.84(20) | 0.84 |
| SMBA-CSFS | 0.83(16) | 0.83(16) | 0.83 | 0.86(20) | 0.86(20) | 0.86 | 0.67(20) | 0.68(20) | 0.67 | 0.8(20) | 0.77(20) | 0.78 | 0.98(15) | 0.98(15) | 0.98 | 0.99(19) | 0.99(19) | 0.99 | 1.0(20) | 1.0(20) | 1.0 | 0.99(20) | 0.98(20) | 0.98 | 0.97(20) | 0.97(20) | 0.97 |
| BSL | 1 | 1 | 1 | 1 | 1 | 1 | 0.74 | 0.74 | 0.74 | 0.92 | 0.92 | 0.92 | 0.93 | 0.93 | 0.93 | 0.8 | 0.8 | 0.8 | 1 | 1 | 1 | 0.98 | 0.98 | 0.98 | 1 | 1 | 1 |