Fuzzy Min-Max Neural Networks for Business Intelligence

Abstract

In this paper the supervised application of fuzzy min-max neural networks to business intelligence is discussed. It utilizes fuzzy sets as pattern classes and builds a fuzzy hyper box for each class in a single pass of the test data. The fuzzy set hyper box is defined by its min point and max point membership functions which are determined by an expansion-contraction process. The best hyper box conforming to the highest memberships is used for the classification of the test data to a particular class.