Table 1.
Parameters | Setting |
---|---|
Epoch | 2000 |
Batch size | 512 |
Leaning rate | 0.0005 |
Optimizer | Adam |
Dropout rate | 0.2 |
Graph convolutional layers | 3 |
Input dimension of the three layers in GNN | N, 4 N, 4 N |
Output dimension of the three layers in GNN | N, 4 N, 4 N |
Fully connected layer hidden unit | 1024, 512 |
Note: N represents the dimension of the initial features.