Open Access
Table 2.
Final parameters used for the training of each model.
| Classification set-up | Batch size | Learning rate | Weight decay | Number of blocks | Scaling |
|---|---|---|---|---|---|
| One-stage | 128 | 1e-5 | 0.005 | 7 | Aggressive arcsinh |
| Two-stage step 1 | 128 | 1e-5 | 0.05 | 4 | Aggressive arcsinh |
| Two-stage step 2 | 128 | 1e-5 | 0.05 | 5 | Aggressive arcsinh |
Notes. The best hyperparameter values found for the models used in the one-stage and two-stage classification set-ups.
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.