Table 1 Alex-Network Architecture: Input Dimension & Number of Kernels
From: Atrial fibrillation classification based on convolutional neural networks
Layer/Model | Alex 1 | Alex 2 | Alex 3 | Alex 4 | Alex 5 | Alex 6 |
|---|---|---|---|---|---|---|
Convolution | (1, 500, 96)a | (1, 500, 48) | (1, 500, 24) | (1, 500, 12) | (1, 500, 6) | (1, 500, 3) |
Pooling | (1, 250, 96) | (1, 250, 48) | (1, 250, 24) | (1, 250, 12) | (1, 250, 6) | (1, 250, 3) |
Convolution | (1, 250, 256) | (1, 250, 128) | (1, 250, 64) | (1, 250, 32) | (1, 250, 16) | (1, 250, 8) |
Pooling | (1, 125, 256) | (1, 125, 128) | (1, 125, 64) | (1, 125, 32) | (1, 125, 16) | (1, 125, 8) |
Convolution | (1, 125, 384) | (1, 125, 192) | (1, 125, 96) | (1, 125, 48) | (1, 125, 24) | (1, 125, 12) |
Convolution | (1, 125, 384) | (1, 125, 192) | (1, 125, 96) | (1, 125, 48) | (1, 125, 24) | (1, 125, 12) |
Convolution | (1, 125, 256) | (1, 125, 128) | (1, 125, 64) | (1, 125, 32) | (1, 125, 16) | (1, 125, 8) |
Pooling | (1, 63, 256) | (1, 63, 128) | (1, 63, 64) | (1, 63, 32) | (1, 63, 16) | (1, 63, 8) |
Fully Connected | (1024) | (1024) | (1024) | (1024) | (1024) | (1024) |
Fully Connected | (1024) | (1024) | (1024) | (1024) | (1024) | (1024) |
Output | (2) | (2) | (2) | (2) | (2) | (2) |