A New Binary Adaptive Elitist Differential Evolution Based Automatic k-Medoids Clustering for Probability Density Functions

Table 9

Comparison of performance of k-means based algorithms and k-medoids based baeDE-CDF in terms of mean of ARI, standard deviation, result of Mann-Whitney test, and ranking by Kruskal-Wallis test.

Data set	Algorithm	Mean	Std.	Result of Mann-Whitney test	Rank by Kruskal-Wallis test
7 normal PDFs	baeDE-CDF	1.000	0.000		142.50
	Chen and Hung	0.442	0.000	+	50.50
	GA-CDF	0.945	0.152	+	135.18
	Vovan and PhamGia	0.469	0.493	+	73.82
Brodatz images	baeDE-CDF	1.000	0.000		162.5
	Chen and Hung	0.316	0.000	+	49.00
	GA-CDF	0.324	0.073	+	56.44
	Vovan and PhamGia	0.772	0.255	+	134.06
UIUC images	baeDE-CDF	1.000	0.000		158.50
	Chen and Hung	0.554	0.000	+	68.50
	GA-CDF	0.541	0.083	+	62.22
	Vovan and PhamGia	0.759	0.368	+	112.78
CUReT images	baeDE-CDF	0.981	0.056		172.7
	Chen and Hung	0.000	0.000	+	26.50
	GA-CDF	0.571	0.084	+	118.41
	Vovan and PhamGia	0.366	0.255	+	84.39
Traffic images	baeDE-CDF	0.909	0.131		173.66
	Chen and Hung	0.000	0.000	+	43.50
	GA-CDF	0.55	0.075	+	127.17
	Vovan and PhamGia	0.044	0.093	+	57.67