Table 1 Teacher model parameters applied.
From: Lightweight bearing fault diagnosis via decoupled distillation and low rank adaptation
Name | Kernel size/stride | Input size | Output size | Activation function | #Parameters |
|---|---|---|---|---|---|
Conv1D_1 | (64, ) / 8 | 1 \(\times\) 1024 | 16 \(\times\) 128 | ReLU | 1072 |
Pooling_1 | 2 / 2 | 16 \(\times\) 128 | 16 \(\times\) 64 | Â | 0 |
Conv1D_2 | (3, ) / 1 | 16 \(\times\) 64 | 32 \(\times\) 64 | ReLU | 1632 |
Pooling_2 | 2 / 2 | 32 \(\times\) 64 | 32 \(\times\) 32 | Â | 0 |
Conv1D_3 | (3, ) / 1 | 32 \(\times\) 32 | 64 \(\times\) 32 | ReLU | 6336 |
Pooling_3 | 2 / 2 | 64 \(\times 32\) | \(64 \times 16\) | Â | 0 |
Conv1D_4 | (3, ) / 1 | 64 \(\times\) 16 | 64 \(\times\) 16 | ReLU | 12480 |
Pooling_4 | 2 / 2 | \(64 \times 16\) | \(64 \times 8\) | Â | 0 |
Conv1D_5 | (3, ) / 1 | 64 \(\times\) 8 | 64 \(\times\) 8 | ReLU | 12480 |
Pooling_5 | 2 / 2 | \(64 \times 8\) | \(64 \times 4\) | Â | 0 |
Conv1D_6 | (3, ) / 1 | 64 \(\times\) 4 | 128 \(\times\) 2 | ReLU | 24960 |
Pooling_6 | 2 / 2 | \(128 \times 2\) | \(128 \times 1\) | Â | 0 |
FC_1 | Â | 128 | 64 | ReLU | 8256 |
FC_2 | Â | 64 | 32 | Â | 2080 |
FC_3 | Â | 32 | 10 | Â | 330 |
Total of trainable parameters | 69626 | ||||
