Intrusion Detection System for Internet of Things Based on Temporal Convolution Neural Network and Efficient Feature Engineering

Table 6

Hyperparameters of deep learning models.

Hyperparameters		Value	Activation function
TCNN	First 1D causal convolution layer	#filters = 64, filter size = 3	ReLU
	Second 1D causal convolution layer	#filters = 128, filter size = 3	ReLU
	Fully connected dense layer	#neurons = 128, dropout = 0. 3	ReLU
	Fully connected dense layer	#neurons = 5	Softmax
CNN	First 1D convolution layer	#filters = 64, filter size = 3	ReLU
	Second 1D convolution layer	#filters = 128, filter size = 3	ReLU
	Fully connected dense layer	#neurons = 128, dropout = 0. 3	ReLU
	Fully connected dense layer	#neurons = 5	Softmax
LSTM	First LSTM layer	#neurons = 20, recurrent dropout = 0.2	ReLU
	Second LSTM layer	#neurons = 20, recurrent dropout = 0.2	ReLU
	Fully connected dense layer	#neurons = 128	ReLU
	Fully connected dense layer	#neurons = 5	Softmax
Optimizer		Adam with learning rate = 0.001	/
Batch size		1024	/
Epochs		15	/