Table 4 Approach parameters.
Layer name | Output size | Parameters | Values | |
|---|---|---|---|---|
KNN-STA-ConvNet | KNN SAM TAM Residual block | 24*10*32 | k | 2 |
[filter, channel; channel] *number of layers | [3*3, 8/16/32]*1 | |||
- | - | |||
- | - | |||
[filter, channel; channel] *number of layers | [3*3, 8/16/32]*1 | |||
ConvLSTM | 24*10*64 | [filter, channel; channel] *number of layers | [3*3, 64]*1 | |
Full connected layer | 256*1 | [layer nodes] *number of layers | [256]*1 | |
10*1 | [10]*1 | |||
Dropout | 0.5 | |||
Batch size | 128 | |||
Learning rate | 0.0001 | |||
Epochs | 50 | |||
Training method | Adam | |||
