Table 5 GCN hyperparameter configuration.
Parameter | Description/value |
|---|---|
Number of GCN layers | 2 convolutional layers |
Hidden dimension | 128 |
Activation function | ReLU |
Dropout rate | 0.3 |
Aggregation type | Mean aggregation |
Normalization | Symmetric degree normalization |
Optimizer | Adam |
Learning rate | 0.01 |
Weight decay (L2 regularization) | 1 × 10⁻⁴ |
Batch size | 64 |
Output function | Softmax |
Initialization | Xavier uniform |
