Training Classifiers under Covariate Shift by Constructing the Maximum Consistent Distribution Subset

Table 3

Comparing ICD2 with SVM and KLIEP.

Data set	SVM	IDC2	KLIEP
Australian	0.787 ± 0.046	0.869 ± 0.035	0.827 ± 0.063
Balance	0.754 ± 0.049	0.823 ± 0.028	0.819 ± 0.055
Breast	0.669 ± 0.059	0.789 ± 0.062	0.783 ± 0.049
Cleveland	0.601 ± 0.068	0.677 ± 0.034	0.671 ± 0.062
Credit	0.790 ± 0.056	0.892 ± 0.068	0.832 ± 0.051
Diabetes	0.802 ± 0.044	0.826 ± 0.039	0.822 ± 0.039
Heart	0.697 ± 0.083	0.818 ± 0.088	0.831 ± 0.062
Ionosphere	0.852 ± 0.046	0.902 ± 0.044	0.899 ± 0.056
Iris	0.901 ± 0.062	0.919 ± 0.056	0.912 ± 0.072
Liver	0.653 ± 0.065	0.745 ± 0.041	0.751 ± 0.053
Page	0.836 ± 0.016	0.932 ± 0.009	0.889 ± 0.022
Sonar	0.727 ± 0.223	0.856 ± 0.102	0.846 ± 0.235
Thyroid	0.859 ± 0.088	0.933 ± 0.128	0.929 ± 0.077
Vehicle	0.665 ± 0.056	0.728 ± 0.032	0.739 ± 0.046
Voting	0.795 ± 0.052	0.929 ± 0.031	0.915 ± 0.038
Waveform21	0.771 ± 0.019	0.832 ± 0.012	0.801 ± 0.026
Waveform40	0.756 ± 0.021	0.815 ± 0.018	0.818 ± 0.025
Wine	0.876 ± 0.096	0.922 ± 0.076	0.916 ± 0.091
wdbc	0.853 ± 0.021	0.935 ± 0.032	0.939 ± 0.012
wpbc	0.681 ± 0.101	0.765 ± 0.086	0.779 ± 0.121