There are a lot of methods to make object classification. It seems that random forest is very used in several domains and is interesting. However, the number of classifiers or classification rules is very huge. Have you an idea about some works which try to reduce the number of classifiers and at the same time they provide similar or better precision?