Table 2:

The Spearman correlation of predicted data values and artificial sample noise rates. The top performing method for each row is bolded; mean ± std.

Dataset	Learning	DVGS	DSHAP	DVRL	LOO	RANDOM
adult	supervised	0.225 ± 0.061	0.130 ± 0.091	0.159 ± 0.074	0.022 ± 0.076	−0.007 ± 0.026
blog	supervised	0.106 ± 0.077	0.086 ± 0.074	0.100 ± 0.344	0.045 ± 0.078	0.011 ± 0.054
cifar10	supervised	0.402 ± 0.081	NA	0.358 ± 0.103	NA	0.000 ± 0.018
cifar10	unsupervised	0.757 ± 0.131	NA	NA	NA	0.003 ± 0.014
lincs (APC>0.5)	unsupervised	0.505 ± 0.018	NA	NA	NA	NA