Table 4. NMI (average and standard deviation) of HC, k-means, SOM, LAC, WSPA, RDCFCE and EM-PCE on eight gene expression datasets.

Dataset	HC	k-means	SOM	LAC	WSPA	RDCFCE	EM-PCE
Breast	0.1636(0)	0.4082(0.0119)	0.4086(0.0013)	0.3446(0.0036)	0.2877(0.0008)	0.4001(0.0025)	0.5408(0.0054)
DLBCLA	0.0295(0)	0.2008(0.0003)	0.2513(0.0006)	0.7958(0.0144)	0.5794(0.0005)	0.3708(0.0000)	0.8525(0.0005)
Leukemia	0.0368(0)	0.8221(0.0002)	0.7160(0.0062)	0.8656(0.0020)	0.6697(0.0016)	0.6706(0.0000)	0.9140(0.0006)
NovartisBPLC	0.5631(0)	0.6541(0.0160)	0.6648(0.0005)	0.8066(0.0072)	0.8885(0.0005)	0.9495(0.0000)	0.9400(0.0000)
Pomeroy2002v2	0.3795(0)	0.6070(0.0008)	0.6055(0.0033)	0.3847(0.0177)	0.7423(0.0012)	0.5842(0.0000)	0.5916(0.0008)
Ramaswamy2001	0.1123(0)	0.4998(0.0009)	0.5309(0.0002)	0.4569(0.0012)	0.6308(0.0000)	0.4949(0.0000)	0.6036(0.0005)
Risinger2003	0.0992(0)	0.2801(0.0034)	0.3918(0.0002)	0.2956(0.0076)	0.3845(0.0010)	0.4712(0.0002)	0.4163(0.0007)
Su2001	0.1357(0)	0.3119(0.0013)	0.3192(0.0005)	0.3604(0.0001)	0.4232(0.0001)	0.3865(0.0006)	0.4157(0.0001)
Average	0.1900(0)	0.4730(0.0044)	0.4860(0.0016)	0.5388(0.0045)	0.5758(0.0007)	0.5410(0.0004)	0.6593(0.0010)

The data in the boldface are the significantly best (or comparable best) results among these comparing methods, and the significance is checked by pairwise t-test at the 95% significance level. The average means the average NMI of each method on eight gene expression datasets.