Table 5.
Performance comparison for independent test samples.
| Dataset | Classifier | # misclassifications (# genes) | |||
| S2N | MMC-RFE(U) | MMC-RFE(O) | SVM-RFE | ||
| Prostate cancer | NNC | 1 (1) | 0 (45) | 0 (22) | 1 (1) |
| NMC | 1 (1) | 0 (2) | 0 (22) | 1 (1) | |
| Leukemia | NNC | 0 (50) | 0 (3) | 0 (3) | 0 (3) |
| NMC | 1 (15) | 0 (54) | 0 (29) | 1 (1) | |
| Breast cancer | NNC | 4 (19) | 3 (91) | 3 (85) | 4 (2) |
| NMC | 4 (1) | 4 (35) | 4 (36) | 1 (2) | |
Minimum number of misclassifications and the number of genes used for Prostate cancer, Leukemia and Breast cancer. The C parameter of SVM-RFE was set to 0.01 for Prostate cancer, and to 0.001 for Leukemia and Breast cancer.