CRND: An Unsupervised Learning Method to Detect Network Anomaly

Table 7

The evaluation results of CRND-200 and K-means method.

Model	Accuracy (%)	Precision (%)	Recall (%)	False-alarm	F1-score	Time (s)
CRND-200	96.13	99.45	91.94	0.42%	0.96	13.10
PCA [37]	86.37	87.52	94.81	—	0.91	13.83
Isolation forest [37]	87.90	92.23	91.07	—	0.88	79
Autoencoder [37]	87.66	91.44	91.64	—	0.92	18.62