Extends neural network analysis to solving multi-classification problems. In these problems, the target output for the network is the probability that the pattern falls into each of several classes, where the number of classes is 3 or greater. These probabilities are then used to assign patterns to one of the target classes. Typical applications include determining the credit classification for a business (excellent, good, fair or poor), and determining which of three or more treatments a patient should receive based upon their physical, clinical and laboratory information. This class signals that network training will minimize the multi-classification cross-entropy error, and that network outputs are the probabilities that the pattern belongs to each of the target classes. These probabilities are scaled to sum to 1.0 using softmax activation.