Improving the Accuracy of Network Intrusion Detection with Causal Machine Learning

Table 11

Comparison of accuracy between SMOTE and CRFS in the UNSW-NB15 dataset.

Feature selection method	Accuracy under different types of cyberattacks (1, 9)
	1	9
SMOTE	0.9357	0.8147
SMOTE (do)	0.9337	0.7499
Percentage	0.9979	0.9205