Estimating a sharp convergence bound for randomized ensembles

Miles E. Lopes

When randomized ensembles such as bagging or random forests are used for binary classification, the prediction error of the ensemble tends to decrease and stabilize as the number o…