Table 3.

Classification error rates on real data. The names in the parenthesis of each data set are the variables used as classification outcome. The best error rates are highlighted with bold font for each column.

Method	Data Sets
	Metabric (Grade)	Glioma (Grade)	Glioma (Site)	Melanoma (Stage)	Melanoma (Met)
PKB-$L_1$	0.274	0.283	0.168	0.304	0.081
PKB-$L_2$	0.304	0.283	0.154	0.307	0.083
Random Forest	0.306	0.302	0.306	0.320	0.136
SVM	0.285	0.292	0.185	0.314	0.083
NPR	0.306	0.298	0.197	0.282	0.110
EasyMKL	0.297	0.302	0.291	0.314	0.100