Research Article
Premature Ventricular Contractions’ Detection Based on Active Learning
Table 5
Each strategy was trained to perform at the level of half of the original data.
| | Random | Least confidence | Margin sampling | Entropy sampling | K-center-greedy | Rep-cluster | Train on the full data set |
| Accuracy | 0.9068 | 0.8621 | 0.8485 | 0.8738 | 0.8648 | 0.8852 | 0.9385 | Sensitivity | 0.786 | 0.8470 | 0.8232 | 0.77267 | 0.81279 | 0.786 | 0.7704 | Specificity | 0.9106 | 0.8626 | 0.8493 | 0.8769 | 0.8663 | 0.8883 | 0.9562 | F 1 score | 0.8237 | 0.8547 | 0.8360 | 0.8215 | 0.8387 | 0.8341 | 0.8533 |
|
|