Table 1.
Deep Neural Networks (DNN) Type 1 layers’ details, including the type of layer, the number of hidden units, the kernel size and the activation function. The kernel size of three was employed in the magnetic differences feature set, while the kernel size of five was used for the combinatorial magnetic differences set.
| Layer | Hidden Units | Kernel Size | Activation Function |
|---|---|---|---|
| Conv1D | 16 | 3 (5) | ReLu |
| Conv1D | 32 | 3 (5) | ReLu |
| Conv1D | 64 | 3 (5) | ReLu |
| Conv1D | 256 | 3 (5) | ReLu |
| MaxPooling1D | - | - | - |
| Dense | 512 | - | Linear |