The most common method of training an artificial neural network. A training set, consisting of examples of input data for which the output is known, is presented to the network, and the network weights are adjusted until the network produces results that are in agreement with the training set. This type of network is often used in prediction and in classification.