Table 3 ResNet50 model parameter settings.
From: Data imbalance in cardiac health diagnostics using CECG-GAN
Block | Layer | Kernel | Padding | Input size | Output size |
|---|---|---|---|---|---|
Preprocess | PCA | / | 256 × 300 × 1 | 256 × 16 × 1 | |
Conv1 | 3 | 1 | 256 × 16 × 1 | 256 × 16 × 64 | |
BatchNorm | / | 256 × 16 × 64 | 256 × 16 × 64 | ||
Relu | / | 256 × 16 × 64 | 256 × 16 × 64 | ||
Layer1 | ResBlock1 | 5 | 2 | 256 × 16 × 64 | 256 × 16 × 128 |
ResBlock2 ~ 8 | 5 | 2 | 256 × 16 × 128 | 256 × 16 × 128 | |
Layer2 | ResBlock1 | 3 | 1 | 256 × 16 × 128 | 256 × 16 × 256 |
ResBlock2 ~ 8 | 3 | 1 | 256 × 16 × 256 | 256 × 16 × 256 | |
Classify | Conv2 | 3 | 1 | 256 × 16 × 256 | 256 × 16 × 256 |
Fc | / | 256 × 1024 | 256 × 5 | ||
Parameter | Value | Parameter | Value | Parameter | Value |
|---|---|---|---|---|---|
Batch size | 256 | Learning rate | 1e-4 | epochs | 100 |
Optimizer | Adam | Loss function | BCELoss | Conv per block | 3 |
Train rate | 0.8 | Â | Â | Â | Â |
