Table 6.

Results for 20-class classification, part I.

Technique	Acc	Acc Avg | Std	F1	F1 Avg | Std	AUC	AUC Avg | Std
FSL proximity-based	44.3–50.1%	47.8% | 2.1%	23.8–26.0	24.7 | 0.8	78.8–84.4	80.8 | 2.5
Softmax-based classification	66.2–68.2%	67.1% | 0.8%	31.9–33.0	32.4 | 0.4	82.4–86.3	84.4 | 1.5
FSL + XGBoost	58.7–66.2%	61.6% | 0.5%	25.3–34.5	29.7 | 3.4	74.2–86.3	79.3 | 3.5
FSL + Random Forest	61.5–66.6%	63.6% | 2.5%	27.1–36.7	30.9 | 3.5	73.6–80.3	77.1 | 2.3
FSL + Decision Tree	45.8–58.8%	51.0% | 4.5%	19.8–30.0	25.0 | 4.1	58.4–60.3	59.5 | 0.6
FSL +KNN − 5 neighbors	58.4–67.2%	61.3% | 3.2%	25.3–36.1	29.6 | 4.1	66.1–70.2	68.2 | 1.6
FSL + KNN − 20 neighbors	61.4–69.9%	64.6% | 2.9%	26.7–36.5	30.9 | 4.0	70.1–76.6	74.1 | 2.3
FSL + SVM with linear kernel	62.9–70.0%	65.3% | 2.5%	28.2–36.7	32.0 | 3.2	77.7–85.8	82.7 | 3.0
FSL + SVM with polynomial kernel	59.7–67.2%	62.9% | 2.8%	23.4–34.0	28.7 | 4.3	74.7–84.9	80.1 | 3.7
FSL + SVM with RBF kernel	63.3–70.6%	65.8% | 2.5%	27.8–37.2	31.4 | 3.9	77.1–82.5	80.5 | 2.2
FSL + SVM with Sigmoid kernel	56.7–65.4%	59.8% | 3.2%	16.6–28.2	23.7 | 4.6	77.0–82.5	80.9 | 2.0