Research Article
Atrial Fibrillation Detection with Low Signal-to-Noise Ratio Data Using Artificial Features and Abstract Features
Table 3
Architecture parameters of the 1-D CNN.
| No. | Layer | Kernel size | Kernel number | Stride | Output size |
| 0 | Input | — | — | — | 1 × 3000 | 1 | Convolution-1 | 5 | 4 | 1 | 4 × 2996 | 2 | Pooling | — | — | 2 | 4 × 1498 | 3 | Convolution-2 | 5 | 8 | 1 | 8 × 1494 | 4 | Pooling | — | — | 2 | 8 × 747 | 5 | Convolution-3 | 7 | 16 | 1 | 16 × 741 | 6 | Pooling | — | — | 2 | 16 × 370 | 7 | Convolution-4 | 7 | 16 | 1 | 16 × 364 | 8 | Pooling | — | — | 2 | 16 × 182 | 9 | Convolution-5 | 9 | 32 | 1 | 32 × 174 | 10 | Pooling | — | — | 2 | 32 × 87 | 11 | Convolution-6 | 11 | 32 | 1 | 32 × 77 | 12 | Pooling | — | — | 2 | 32 × 38 | 13 | FC | — | — | — | 1 × 38 |
|
|