Table 11 The hyperparameter grid used in the hyperparameter optimization
Parameter | Values |
|---|---|
Batch size | \(\{8, 16, 32\}\) |
Dropout rate | \(\{0.1, 0.3, 0.5, 0.7\}\) |
Epochs | \(\{2..20\}\) |
Gamma | \(\{0.0, 0.1, 0.3, 0.5\}\) |
Layer dimension | \(\{32, 64, 128, 256, 512, 1024\}\) |
Learning rate | \(\{0.001, 0.01\}\) |
Margin | \(\{0.2, 0.5, 1\}\) |
Weight decay | \(\{0.0001, 0.001, 0.01, 0.05, 0.1\}\) |
PCA variance threshold | \(\{0.9, 0.95, 0.975, 0.99\}\) |