An Efficient Communication Intrusion Detection Scheme in AMI Combining Feature Dimensionality Reduction and Improved LSTM

Table 10

The distribution of various types of intrusion attacks in the NSL-KDD dataset.

Dataset class	Attack class	Size	Distribution (%)
Training dataset	Normal	67343	53.46
	DoS	45927	36.46
	Probe	11656	9.25
	R2L	995	0.79
	U2R	52	0.04
Testing dataset	Normal	9711	43.07
	DoS	7458	33.08
	Probe	2421	10.74
	R2L	2754	12.22
	U2R	200	0.89