Table 2 Perforamcne measures of various models
LMNet | CoroNet | Deep GRU-CNN | CVDNet | Proposed Model | |
|---|---|---|---|---|---|
Total Parameters | 8,92,226 | 2,31,66,674 | 1,65,26,978 | 53,17,154 | 83,307 |
Trainable Parameters | 8,92,162 | 2,31,12,146 | 8,23,490 | 53,15,618 | 81,647 |
Non-trainable parameters | 64 | 54,528 | 1,57,03,488 | 1536 | 1660 |
Size in MB | 10.89 | 278.36 | 69.55 | 64.13 | 1.18 |
Training Accuracy | 95.69% | 99.18% | 97.4% | 98.88% | 99.47% |
Training loss | 0.1377 | 0.0286 | 0.0705 | 0.0296 | 0.0196 |
Training AUC | 0.9893 | 0.9993 | 0.9973 | 0.9993 | 0.9995 |
Test Accuracy | 92.27% | 97.76% | 93.48% | 97.78% | 98.91% |
Test loss | 0.2999 | 0.0719 | 0.1614 | 0.0713 | 0.0397 |
Test AUC | 0.9684 | 0.9959 | 0.985 | 0.9962 | 0.9983 |
Sensitivity | 0.7944 | 0.9545 | 0.9727 | 0.8939 | 0.9819 |
Specificity | 0.9848 | 0.9819 | 0.9508 | 0.9969 | 0.9949 |
Precision | 0.9289 | 0.9112 | 0.75 | 0.9851 | 0.9754 |
F1-score | 0.8564 | 0.9323 | 0.847 | 0.9323 | 0.9786 |