We compare the performance of three types of neural network-based ense
mble techniques to that of a single neural network. The ensemble algor
ithms are two versions of boosting and committees of neural networks'
trained independently. For each of the four algorithms, we experimenta
lly determine the test and training error curves in an optical charact
er recognition (OCR) problem as both a function of training set size a
nd computational cost using three architectures. We show that a single
machine is best for small training set size while for large training
set size some version of boosting is best. However, for a given comput
ational cost, boosting is always best. Furthermore, we show a surprisi
ng result for the original boosting algorithm: namely, that as the tra
ining set size increases, the training error decreases until it asympt
otes to the test error rate. This has potential implications in the se
arch for better training algorithms.