A Method of Extracting Pattern Recognition Rules Based on Multiple Precision Fuzzy Partition

A method is proposed to extract an appropriate fuzzy rule set automatically for classification problems that have several fuzzy eigenvariables.Multiple fuzzy rule tables are generated firstly by multiple precision partition of fuzzy space,and then an optimized set of fuzzy rules for pattern recognition can be obtained by the clone selection algorithm according to the principle of artificial immunology.Experiment shows that the rule set selected with this method has less rules,and higher ratio of correct classification.