Abstract
Neural networks are suggested for learning a map from $d$-dimensional sampleswith any underlying dependence structure to multivariate uniformity in $d'$dimensions. This map, termed DecoupleNet, is used for dependence modelassessment and selection. If the data-generating dependence model was known,and if it was among the few analytically tractable ones, one suchtransformation for $d'=d$ is Rosenblatt's transform. DecoupleNets only requirean available sample and are applicable to $d'<d$, in particular $d'=2$. Thisallows for simpler model assessment and selection without loss of information,both numerically and, because $d'=2$, graphically. Through simulation studiesbased on data from various copulas, the feasibility and validity of this novelapproach is demonstrated. Applications to real world data illustrate itsusefulness for model assessment and selection.