Skip to main content
. 2023 Feb 9;9(2):e13520. doi: 10.1016/j.heliyon.2023.e13520

Table 3.

The structure of HyDL-IDS.

Layer (type) Output Shape Param #
conv1d (Conv1D) (None, 57, 16) 64
batch_normalization (BatchNo) (None, 57, 16) 64
max_pooling1d (MaxPooling1D) (None, 28, 16) 0
dropout (Dropout) (None, 28, 16) 0
conv1d_1 (Conv1D) (None, 24, 32) 2592
batch_normalization_1 (BatchNo) (None, 24, 32) 128
max_pooling1d_1 (MaxPooling1D) (None, 12, 32) 0
dropout_1 (Dropout) (None, 12, 32) 0
conv1d_2 (Conv1D) (None, 8, 64) 10304
batch_normalization_2 (BatchNo) (None, 8, 64) 256
max_pooling1d_2 (MaxPooling1D) (None, 4, 64) 0
dropout_2 (Dropout) (None, 4, 64) 0
conv1d_3 (Conv1D) (None, 2, 128) 24704
batch_normalization_3 (BatchNo) (None, 2, 128) 512
max_pooling1d_3 (MaxPooling1D) (None, 1, 128) 0
dropout_3 (Dropout) (None, 1, 128) 0
lstm (LSTM) (None, 256) 394240
flatten (Flatten) (None, 256) 0
dense_6 (Dense) (None, 256) 65792
dropout_4 (Dropout) (None, 256) 0
dense_7 (Dense) (None, 10) 2750
Total params: 501,226
Trainable params: 500,746
Nontrainable params: 480