Table 1 The detailed hyperparameter settings and output dimensions of each layer of the SE-TCN
From: Hybrid-FHR: a multi-modal AI approach for automated fetal acidosis diagnosis
Layer | Output shape | Kernel size | Stride | Padding | Dilation factor | Activation |
|---|---|---|---|---|---|---|
Input | B × 7200 × 1 | – |  | – | – | – |
MDSC | B × 7200 × 64 | [1, 3, 5, 7] | 1 | Same | 1 | ReLu |
SE-TCNBlock1 | B × 2400 × 64 | 15 | 3 | Casual | 2 | ReLu |
SE-TCNBlock2 | B × 2400 × 64 | 15 | 1 | Casual | 4 | ReLu |
SE-TCNBlock3 | B × 800 × 128 | 15 | 3 | Casual | 8 | ReLu |
SE-TCNBlock4 | B × 800 × 128 | 15 | 1 | Casual | 16 | ReLu |
SE-TCNBlock5 | B × 400 × 256 | 15 | 2 | Casual | 32 | ReLu |