Extracting fuzzy rules from pattern classification neural networks

A technique for generating fuzzy rules for pattern classification is discussed. First, separation hyperplanes for classes are extracted from a trained neural network. Then, for each class, convex existence regions in the input space are approximated by shifting these hyperplanes in parallel using the training data set for the classes. Using fuzzy rules defined for each class, input data are directly classified without use of the neural network. This method is applied to a number recognition system as well as to a blood cell classification system; and their performance is compared with that obtained with neural networks