Hidden Service Website Response Fingerprinting Attacks Based on Response Time Feature

Table 4

Performance comparison of different algorithms in the classifiers.

Classifier	Algorithm	Accuracy (%)	Test time (hour)	Parameters	Features	Memory (GB)	Page
WRFP	Extremely randomized tree	93.63	0.43	n = 213	87	8	10,000
	Random forest	91.82	0.87	n = 487	87	8	10,000
K-FP	Random forest + Nearest neighbors	92.35	1.14	K = 5, n = 379	186	65	10,000
K-NN	K-nearest neighbors	88.14	1.65	K = 5	1225	10	10,000
		88.06	4.88	K = 5	3736	16	10,000
CUMUL	SVM-RBF	90.78	0.55	C = 4096	150	30	10,000

Test time refers to the instance feature extraction time + page classification time. Memory denotes the minimum required memory size.