Wrapping (feature selection)
Suppose we have some learning problem where the input domain has a large number of features. Filtering is the process of selecting features for a learning algorithm but using that algorithm to evaluate how it performs on subsets of the data.