In
this context, one of their possible applications is in the domain of variable selection, where the task is to find the optimal subset of descriptors/predictors, e.g. for improving the predictive ability of a regression
or classification model or for biomarker discovery.