Table 8.
The classification testing accuracy (%) for the five classifiers used in MB-AI-His trained using spatial-time-frequency DL features extracted from the three CNNs.
| Binary Classification Level | |||||
|---|---|---|---|---|---|
| Features | Linear-SVM | Cubic-SVM | k-NN | LDA | ESD |
| Spatial-Time-Frequency-ResNet-50 | 100 | 100 | 98.2 | 100 | 99 |
| Spatial-Time-Frequency-MobileNet | 98.2 | 98.4 | 98.4 | 98 | 98.2 |
| Spatial-Time-Frequency-DenseNet-201 | 98.4 | 98.2 | 98.4 | 98.2 | 99.2 |
| Multi-Class Classification Level | |||||
| Spatial-Time-Frequency-ResNet-50 | 94.32 | 95.22 | 89.48 | 96.66 | 95.36 |
| Spatial-Time-Frequency-MobileNet | 93.76 | 94.18 | 90.14 | 94.32 | 93.66 |
| Spatial-Time-Frequency-DenseNet-201 | 95.74 | 97.16 | 95.08 | 98.46 | 97.42 |