A Method of Selecting Fuzzy Rules for Pattern Identification Based on Multi-Precision Fuzzy Partitions
Qing Ye, Yan Zhao, Chai Chang, Chen Zhong · 2007
It's important to extract an appropriate fuzzy rule set for multi-classification problems that have fuzzy variables. This paper proposes a new method to make the fuzzy partitions with multi-precision firstly, then produces multiple fuzzy rule tables, makes optimization to obtain a group of elite fuzzy rules by clone selection algorithm. The simulated experiment shows that the method has the performances of fewer fuzzy rules, higher classification correctness and better plasticity than that of single fuzzy partition.