Anomaly Detection Collaborating Adaptive CEEMDAN Feature Exploitation with Intelligent Optimizing Classification for IIoT Sparse Data

Table 4

Detection efficiency comparisons between traditional ABC-OCSVM and PSO-OCSVM anomaly detection classifiers.

Attack power	ABC-OCSVM			IABC-OCSVM
	Training accuracy	Test accuracy	Test time	Training accuracy	Test accuracy	Test time
1	96.0%	86.5%	0.0079 s	96.0%	86.5%	0.0076 s
2	93.5%	88.0%	0.0079 s	92.0%	88.5%	0.0080 s
3	91.0%	90.0%	0.0090 s	91.0%	90.5%	0.0081 s
4	96.5%	87.0%	0.0076 s	95.0%	88.0%	0.0079 s
5	98.5%	93.5%	0.0086 s	98.5%	95.5%	0.0089 s
Average	95.10%	89.00%	0.0082 s	94.50%	89.80%	0.0081 s