Flow Correlation Degree Optimization Driven Random Forest for Detecting DDoS Attacks in Cloud Computing

Table 1

Comparison results of four algorithm detection evaluation criteria with changing numbers of normal training samples.

		Sample numbers
		30	50	70	90
GAORF; (%)	accuracy	98.57	99.52	100	100
	MR	0	0	0	0
	FR	2.72	0.91	0.0	0.0
nu-SVM; (%)	accuracy	93.33	85.24	99.05	100
	MR	0	0	0	0
	FR	12.72	28.18	1.81	0
C-SVM; (%)	accuracy	91.90	100	100	100
	MR	0	0	0	0
	FR	15.45	0	0	0
one-class-SVM; (%)	accuracy	37.62	38.10	40.95	45.71
	MR	21.00	21.00	21.00	21.00
	FR	100	99.09	93.64	84.55