Free Access
Table 4.
Classifying GCs from the combined image data sets of Virgo and Fornax.
| Method | TPR | FPR | FDR | AUC ROC |
|---|---|---|---|---|
| Nearest Neighbour | 0.842±0.005 | 0.033±0.002 | 0.122±0.006 | − |
| 12 Nearest Neighbours | 0.847±0.008 | 0.023±0.001 | 0.088±0.004 | 0.975±0.002 |
| Random Forest | 0.838±0.005 | 0.026±0.001 | 0.100±0.005 | 0.981±0.001 |
| Convolutional Neural Network (CNN) | 0.929±0.015 | 0.019±0.003 | 0.068±0.011 | 0.994±0.001 |
| CNN + Nearest Neighbour | 0.912±0.006 | 0.026±0.001 | 0.094±0.004 | − |
| CNN + 12 Nearest Neighbours | 0.927±0.007 | 0.017±0.001 | 0.061±0.004 | 0.990±0.001 |
Notes. The reported results are averages over ten random splits (train, validation and test). Uncertainties are given as standard deviations and the used decision threshold is 0.5.
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.