Rule weight generation for a fuzzy classification system based on fuzzy clustering methods

The authors propose a method of automated generation of a fuzzy classification system consisting of a fuzzification unit, a rule evaluation unit and a rule weighting unit based on fuzzy clustering algorithms. All three parts of the classification system are generated automatically using training data, expert knowledge expressed in natural language may be integrated into the system, if available. The authors focus on the topic of the generation of rule weights to be used in the rule weighting unit. Several methods are discussed and tested with an example from industrial quality control