Network Intrusion Anomaly Detection Model Based on Multiclassifier Fusion Technology

Table 4

Anomaly detection algorithm performance comparison (anomaly ratio of 0.1).

Dataset	Performance metrics	LODA	AE	PCA	HBOS	iForest	WV	WMV	NB
CICIDS2017	AUC	0.629	0.643	0.683	0.513	0.753	0.789	0.819	0.842
	RECALL	0.389	0.421	0.569	0.492	0.628	0.853	0.853	0.896
	Time (seconds)	0.441	6.649	0.484	0.197	7.391	3.724	3.621	3.604
UNSW-NB 15	AUC	0.721	0.691	0.783	0.689	0.927	0.919	0.924	0.943
	RECALL	0.752	0.803	0.902	0.762	0.932	0.934	0.951	0.958
	Time (seconds)	0.529	8.236	0.692	0.384	9.242	4.045	4.578	4.248
KDDCUP 99	AUC	0.928	0.956	0.931	0.928	0.954	0.894	0.927	0.961
	RECALL	0.952	0.969	0.950	0.955	0.985	0.961	0.985	0.993
	Time (seconds)	0.502	7.923	0.669	0.328	8.927	3.928	4.029	4.293