Table 4 The details of 2D CNN structure
From: Automatic seizure detection using three-dimensional CNN based on multi-channel EEG
Layer | Hidden Layer | Related parameters (kernel, kernel size, stride, dropout) | ||
|---|---|---|---|---|
1 | Conv2D + LeakyReLU+BN | 32 | 5*5 | 1 |
2 | Max Pooling | Â | 3*3 | 2 |
3 | Conv2D + LeakyReLU+BN | 64 | 3*3 | 1 |
4 | Max Pooling | Â | 2*2 | 2 |
5 | Conv2D + LeakyReLU+BN | 128 | 3*3 | 1 |
6 | Max Pooling | Â | 2*2 | 2 |
7 | Conv2D + LeakyReLU+BN | 256 | 3*3 | 1 |
8 | Max Pooling | Â | 2*2 | 2 |
9 | Conv2D + LeakyReLU+BN | 256 | 3*3 | 1 |
10 | Max Pooling | Â | 2*2 | 2 |
11 | Fully connected | 2048 | ||
Dropout | 0.5 | |||
12 | Fully connected | 1024 | ||
Dropout | 0.5 | |||
Softmax | ||||