Table 13.
Testing precision and F-score using the untrained VGG19 network architecture
| Training epochs (%) | 25 (%) | 50 (%) | 75 (%) | 100 (%) | ||||
|---|---|---|---|---|---|---|---|---|
| Threshold | Precision | F score | Precision | F score | Precision | F score | Precision | F score |
| 60 | 83.88 | 80.26 | 84.12 | 80.98 | 85 | 81.31 | 86.12 | 82.96 |
| 70 | 78.03 | 74.61 | 81.02 | 77.76 | 82.98 | 79.75 | 84.13 | 81.43 |
| 80 | 55.55 | 51.13 | 75.11 | 71.88 | 79.46 | 76.45 | 79.67 | 76.34 |
| 90 | – | – | 60.4 | 57.4 | 70.87 | 67.45 | 78.5 | 75.16 |
The network was trained, validated, and tested using the dataset without repeating any K-complex. Threshold is the positive overlap threshold