Table 2.
The shapes of the proposed model.
| Number of Layers | Layer Type | Numbers of Input Channels/Output Channels |
|---|---|---|
| 1 | Input (shape:1, 384, 32) | |
| 2 | conv_1 (Conv2d) | 1/25 (kernel size: 5 × 1) |
| 3 | droputout1 (Dropout=0.25) | 1/25 |
| 4 | conv_2 (Conv2d) | 25/25 (kernel size: 1 × 3, stride = (1,2)) |
| 5 | bn1 (BatchNorm2d) | 25 |
| 6 | pool1 (MaxPool2d (2,1)) | 25/25 |
| 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 |
conv_3 (Conv2d) droputout2 (Dropout = 0.25) conv_4 (Conv2d) bn2 (BatchNorm2d)pool2 (MaxPool2d (2,1)) conv_5 (Conv2d) droputout3 (Dropout = 0.25) conv_6 (Conv2d) bn3 (BatchNorm2d) pool3 (MaxPool2d (2,1)) conv_7 (Conv2d) droputout4 (Dropout = 0.25) conv_8 (Conv2d) bn4 (BatchNorm2d) flatten (Flatten layer) Linear1 (Linear) Droputout5 (Dropout = 0.5) Linear2 (Linear) |
25/50 (kernel size: 5 × 1) 25/50 50/50 (kernel size: 1 × 3, stride = (1,2)) 50 50 50/100 (kernel size: 5 × 1) 50/100 100/100 (kernel size: 1 × 3, stride = (1,2)) 100 100 100/200 (kernel size: 5 × 1) 100/200 200/200 (kernel size: 1 × 3) 200 Shape: 128 × 8000 8000/256 256/2 (binary classification task, number of classes = 2) |