A neuro-fuzzy-genetic classifier for technical applications

Abstract

The paper presents an approach that combines artificial neural networks with fuzzy logic to form a neuro-fuzzy classifier. The proposed system has a feedforward network-like structure that mirrors fuzzy rules. The proposed system is able to learn and to generalize gained knowledge (it comes from the network-like structure) as well as to explain the decisions it makes. Its learning abilities are strengthened by applying a genetic algorithm as a technique of global optimization. The proposed neuro-fuzzy classifier has been successfully applied to the glass identification problem in forensic science.