Table 7.

Results for 20-class classification, part II.

Technique	Specificity	Specificity Avg | Std	Sensitivity	Sensitivity Avg | Std
FSL proximity-based	24.9–26.8%	25.6% | 0.6%	27.7–29.7%	28.5% | 0.7%
Softmax-based classification	36.3–39.7%	37.6% | 1.2%	32.2–33.1%	32.6% | 0.3%
FSL + XGBoost	23.8–31.6%	27.7% | 2.6%	25.3–34.5%	29.7% | 3.4%
FSL + Random Forest	26.6–32.3%	28.4% | 2.2%	27.1–36.7%	31.0% | 3.6%
FSL + Decision Tree	19.4–28.5%	24.8% | 3.7%	19.9–30.0%	25.1% | 4.2%
FSL +KNN − 5 neighbors	24.0–33.7%	28.3% | 3.3%	25.3–36.1%	29.6% | 4.1%
FSL + KNN − 20 neighbors	24.9–31.1%	28.2% | 2.0%	26.7–36.5%	30.9% | 4.0%
FSL + SVM with linear kernel	25.6–34.1%	28.8% | 3.1%	28.2–36.7%	32.0% | 3.2%
FSL + SVM with polynomial kernel	26.0–33.0%	29.1% | 2.4%	23.4–34.0%	28.7% | 4.3%
FSL + SVM with RBF kernel	25.5–31.1%	28.9% | 2.7%	27.8–37.2%	31.4% | 3.9%
FSL + SVM with Sigmoid kernel	15.5–25.4%	20.8% | 3.4%	16.6–28.2%	23.7% | 4.6%