Table 6.
Hyperparameter | MLP | CNN |
---|---|---|
Number of fully connected layers | 1, 2, 3 | 1 |
Number of convolution blocks | - | 1, 2 |
Kernel size | - | 3, 5, 7 |
Filter number | - | 16, 32, 64 |
Stride | - | 1 |
Pooling size | - | 2 |
Hidden units | 16, 32, 64 | |
Dropout rate | 0.1, 0.3, 0.5 | |
Learning rate | 0.001, 0.0005, 0.0001 |