Table 2. Classification results on AlexNet DCNN features. Best values are shown in bold.
| Classifier | Performance measures | ||||
|---|---|---|---|---|---|
| Sensitivity (%) | Specificity (%) | FNR (%) | Accuracy (%) | F-score | |
| Cubic SVM | 91.9 | 90.5 | 8.0 | 91.2 | 0.913 |
| Linear discriminant | 72.6 | 69.8 | 27.3 | 71.2 | 0.716 |
| Linear SVM | 86.1 | 85.8 | 12.6 | 86.4 | 0.860 |
| Quadratic SVM | 91.2 | 90.1 | 8.7 | 90.7 | 0.907 |
| Fine KNN | 89.3 | 86.9 | 10.6 | 88.1 | 0.882 |
| Medium KNN | 92.9 | 83.8 | 7.0 | 88.4 | 0.889 |
| Cubic KNN | 93.1 | 83.6 | 6.8 | 88.4 | 0.889 |
| Weighted KNN | 91.9 | 85.6 | 8.0 | 88.9 | 0.891 |
| Subspace discriminant | 74.1 | 66.4 | 25.8 | 70.3 | 0.714 |
| Subspace KNN | 89.7 | 87.2 | 10.2 | 88.5 | 0.886 |