Table 6.
Comparison of classification accuracies across different feature selection methods in cross-validation set
| Classifier | RIMONE | SLO images | ||||
|---|---|---|---|---|---|---|
| wrap-AUC | wrap-LDA | wrap-QDA | wrap-AUC | wrap-LDA | wrap-QDA | |
| Twin SVM | 96.3 % | 90.0 % | 78.8 % | 94.1 % | 84.3 % | 78.4 % |
| Linear SVM | 95.0 % | 90.0 % | 81.3 % | 94.1 % | 84.3 % | 78.4 % |
| Polynomial SVM | 95.0 % | 90.0 % | 81.3 % | 94.1 % | 84.3 % | 78.4 % |
| RBF SVM | 90.0 % | 87.5 % | 82.5 % | 82.3 % | 78.4 % | 82.3 % |
| Sigmoid SVM | 78.8 % | 92.5 % | 77.5 % | 78.4 % | 74.5 % | 78.4 % |
| LDA | 95.0 % | 88.8 % | 80.0 % | 90.5 % | 82.4 % | 78.4 % |
| QDA | 85.0 % | 81.3 % | 86.3 % | 78.4 % | 68.6 % | 82.4 % |