Table 5.
Results for five-class classification, part II.
| Technique | Specificity | Specificity Avg | Std | Sensitivity | Sensitivity Avg | Std |
|---|---|---|---|---|
| FSL proximity-based | 60.2–66.4% | 63.2% | 2.1% | 62.7–68.1% | 65.9% | 2.0% |
| Softmax-based classification | 68.3–70.6% | 69.5% | 0.8% | 65.9–69.1% | 67.1% | 1.1% |
| FSL + XGBoost | 65.7–68.0% | 66.9% | 0.9% | 66.9–70.8% | 68.4% | 1.7% |
| FSL + Random Forest | 66.8–68.3% | 67.8% | 0.6% | 66.8–69.9% | 68.4% | 1.2% |
| FSL + Decision Tree | 57.1–59.9% | 59.0% | 1.0% | 58.8–63.4% | 61.3% | 1.5% |
| FSL + KNN − 5 neighbors | 64.9–70.0% | 67.6% | 1.8% | 65.9–71.1% | 68.4% | 2.1% |
| FSL + KNN − 20 neighbors | 70.2–72.9% | 71.9% | 1.0% | 68.2–71.6% | 69.6% | 1.2% |
| FSL + SVM with linear kernel | 69.8–74.5% | 72.4% | 1.6% | 69.9–73.1% | 71.7% | 1.1% |
| FSL + SVM with polynomial kernel | 68.3–75.5% | 72.4% | 2.5% | 64.7–69.2% | 66.6% | 1.6% |
| FSL + SVM with RBF kernel | 70.8–75.5% | 73.5% | 1.8% | 69.0–71.8% | 70.6% | 1.0% |
| FSL + SVM with Sigmoid kernel | 56.1–70.8% | 65.4% | 5.1% | 53.1–65.0% | 62.1% | 4.5% |