Table 6 EEGBaseNet and fNIRSBaseNet structures.
Network | Layer | Input | Operation | Filter size | Padding | Activation | Output |
|---|---|---|---|---|---|---|---|
EEG Base Net | Conv1 | 32 * 2560 | Conv1D | F1 * 63 | 31 | Sigmoid | F1 * 2560 |
BathNorm | |||||||
Conv2 | F1 * 2560 | Conv1D | F2 * 32 | / | Sigmoid | F2 * 2529 | |
BathNorm | |||||||
Pooling1 | F2 * 2529 | Avgpooling | 4 | / | / | F2 * 632 | |
Conv3 | F2 * 632 | SeprateConv1D | F2 * 15 | 7 | Sigmoid | F2 * 632 | |
BathNorm | |||||||
Pooling2 | F2 * 632 | Avgpooling | 2 | / | / | F2 * 316 | |
Flatten | F2 * 316 | / | / | / | / | (F2 * 316) | |
Classifier | (F2 * 316) | Linear | (F2 * 246) * 2 | / | / | 2 | |
fNIRS Base Net | Conv1 | 90 * 110 | Conv1D | F1 * 63 | 31 | Sigmoid | F1 * 110 |
BathNorm | |||||||
Conv2 | F1 * 110 | Conv1D | F2 * 90 | 45 | Sigmoid | F2 * 110 | |
BathNorm | |||||||
Pooling1 | F2 * 110 | Avgpooling | 4 | / | / | F2 * 27 | |
Conv3 | F2 * 27 | SeprateConv1D | F2 * 15 | 7 | Sigmoid | F2 * 17 | |
BathNorm | |||||||
Pooling2 | F2 * 27 | Avgpooling | 2 | / | / | F2 * 13 | |
Flatten | F2 * 13 | / | / | / | / | (F2 * 13) | |
Classifier | (F2 * 13) | Linear | (F2 * 13) * 2 | / | / | 2 |
