Neural Expert System with two Engines of Rough Set

Abstract

A new method for preprocessing input to a neural network based on rough sets analysis is proposed. The method is used to construct two engines of rough sets within the frame of a neural expert system. As an example, a neural expert system for medical diagnosis was built using the two engines. The resulting model is evaluated by a fivefold cross-validation technique. The obtained result shows consistency of the given data and the classification given by experts