Table 5
Performance of the SS-3D-Clump model trained on datasets from three regions when applied to the other regions.
| Evaluation | HDR-trained model | Retraining | MDR-trained model | Retraining | LDR-trained model | Retraining | |||
|---|---|---|---|---|---|---|---|---|---|
| R | MDR data | 0.963 | 0.980 | HDR data | 0.961 | 0.950 | HDR data | 0.947 | 0.949 |
| P | 0.720 | 0.987 | 0.910 | 0.918 | 0.947 | 0.947 | |||
| F1 | 0.824 | 0.983 | 0.935 | 0.933 | 0.947 | 0.948 | |||
| Acc | 0.792 | 0.978 | 0.904 | 0.903 | 0.924 | 0.925 | |||
| R | LDR data | 0.966 | 0.973 | LDR data | 0.980 | 0.965 | MDR data | 0.951 | 0.993 |
| P | 0.918 | 0.910 | 0.869 | 0.906 | 0.770 | 0.962 | |||
| F1 | 0.941 | 0.941 | 0.921 | 0.934 | 0.851 | 0.977 | |||
| Acc | 0.940 | 0.939 | 0.917 | 0.902 | 0.832 | 0.969 | |||
Notes. In the table, the column ‘HDR trained model’ represents the performance of the model trained on the HDR dataset when applied to the MDR and LDR datasets. The adjacent ‘Training’ column represents the performance after fine-tuning the model trained on the HDR dataset and applying it separately to the MDR and LDR datasets. The subsequent columns follow a similar pattern.
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.