Free Access
Table 11.
Testing accuracy of the four selected algorithms for the three-class classification of 4FGL-DR2 sources.
| Algorithm | Parameters | Testing | Std. Dev. |
|---|---|---|---|
| accuracy | |||
| RF | 50 trees, max depth 6 | 92.91 | 0.66 |
| RF_O | 92.83 | 0.63 | |
| BDT | 100 trees, max depth 2 | 92.51 | 0.67 |
| BDT_O | 92.27 | 0.67 | |
| NN | 600 epochs, 16 neurons, LBFGS | 91.86 | 0.72 |
| NN_O | 90.26 | 0.83 | |
| LR | 1000 iterations, LBFGS solver | 92.63 | 0.67 |
| LR_O | 92.22 | 0.69 | |
Notes. “_O” denotes training with oversampling.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.