Generating fuzzy rules for a neural fuzzy classifier
Chihwen Li, Chwan-Hwa Wu · 1994
It is difficult to design a classifier when overlap problems occur between the decision regions of different classes. A top-down learning procedure is described which trains a neural fuzzy (NF) system from global to local views of the overlap decision region, and generates nested IF-THEN rules. With these rules, the NF system can correctly separate similar classes within the overlap decision region. Two operational examples of the NF system are given.>