ABSTRACT
Given a set of trained individual learners, rather than combining all of them, ensemble pruning tries to select a subset of individual learners to comprise the ensemble.
An apparent advantage of ensemble pruning is to obtain ensembles with smaller sizes; this reduces the storage resources required for storing the ensembles and the computational resources required for calculating outputs of individual learners, and thus improves efficiency. There is another benefit, that is, the generalization performance of the pruned ensemble may be even better than the ensemble consisting of all the given individual learners.
