Table 1 The setting of parameters for the developed network.
Type of hyper-parameter | Value |
|---|---|
Epoch | 300 |
Batch size | 1 |
Initial learning rate | 1e − 4 |
Learning rate decay strategy | Cosine annealing |
Optimizer | Adam |
Loss function | Dice loss |
Supervisory loss weights | 0.7 |
Cross-validation split | 0.2 |
Patch embedding dim | 48 |
Window size | 7 |
