Pattern Classification

• Learning Curves33• Before training starts, the error on the training set is high;as the learning proceeds, error becomes smaller• Error per pattern depends on the amount of training dataand the expressive power (such as the number ofweights) in the network• Average error on an independent test set is always higherthan on the training set, and it can decrease as well asincrease• A validation set is used in order to decide when to stoptraining ; we do not want to overfit the network anddecrease the power of the classifier’s generalization“Stop training when the error on the validation set isminimum”Pattern Classification, Chapter 6