Table 10 Using full convolutional basis with different networks for feature extraction and fusion classification performance
Strategie | Classifer | |||
|---|---|---|---|---|
A1 | A2 | A3 | A4 | |
Sigmoid-SVM | 44.20/63.33 | 62.32/73.17 | 54.33/73.00 | 39.13/36.00 |
Rbf-SVM | 62.32/98.67 | 58.70/91.83 | 55.80/87.00 | 39.86/36.33 |
Poly-SVM | 58.70/99.38 | 57.97/94.67 | 54.35/92.00 | 44.93/43.67 |
Linear-SVM | 39.13/36.00 | 55.07/49.33 | 39.13/36.00 | 39.13/36.00 |
RF | 59.42/100.00 | 64.49/100.00 | 59.42/89.67 | 42.75/73.33 |
KNN | 41.30/100.00 | 57.25/97.00 | 55.07/92.00 | 32.61/62.00 |
Strategie | Classifer | |||
|---|---|---|---|---|
A1 ∪ A2 | A1 ∪ A3 | A2 ∪ A3 | A1 ∪ A2 ∪ A3 | |
Sigmoid-SVM | 57.25/79.67 | 58.70/78.67 | 60.87/83.33 | 57.25/84.00 |
Rbf-SVM | 60.87/92.67 | 55.80/89.67 | 61.59/94.33 | 63.04/93.67 |
Poly-SVM | 60.87/94.67 | 57.25/93.00 | 64.49/95.00 | 63.04/96.00 |
Linear-SVM | 39.13/36.00 | 39.13/36.00 | 39.13/37.00 | 41.30/38.33 |
RF | 61.59/95.00 | 59.42/92.67 | 39.13/36.00 | 61.59/95.00 |
KNN | 50.00/97.33 | 53.62/95.00 | 57.25/96.67 | 50.00/96.67 |