From: Network traffic classification model based on attention mechanism and spatiotemporal features
Parameter | Values |
---|---|
Learning rate | \({{10}^{-2},{ 10}^{-3},10}^{-4}\) |
Number of training (12 classifications) | \({3.5 \times 10}^{5}\) |
Number of training (20 classifications) | \({6.5 \times 10}^{5}\) |
Batch (12 classifications) | 64 |
Batch (20 classifications) | 256 |
Time stamp of LSTM | 28 |
N_inputs (the number of input images of LSTM) |  >  = 1 |
N_classes (the number of network traffic classes) |  >  = 2 |
N_hidden_units (the number of hidden neurons of LSTM) | 28 |
batch_size (batch size of LSTM) | 64 |
Epochs (the number of iterations) |  >  = 300 |
Dropout | 0.3, 0.5, 0.7 |