A general model for finite-sample effects in training and testing of competing classifiers | IEEE Journals & Magazine | IEEE Xplore