Table 2.
Investigated parameters for the training process of the FCNN (see Fig. 5) for the system listed in Table 1.
| α | 0.01, 0.001, 0.0001, 0.00001 |
|---|---|
| f | 5, 10, 20, 40, 80, 160 |
| Nbatch | 64, 128, 256, 512 |
| Nepoch | 400 |
Notes. We vary the learning rate, α, the size of the hidden layers by a factor f, and the size of the mini batches, Nbatch. Furthermore, 400 training epochs (Nepoch) are sufficient given that the minimum loss of the validation set is typically reached around 200 training iterations.
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.