作者:
关键词: Fuzzy rule 、 Fuzzy classification 、 Fuzzy set operations 、 Defuzzification 、 Membership function 、 Fuzzy number 、 Type-2 fuzzy sets and systems 、 Mathematics 、 Neuro-fuzzy 、 Artificial intelligence 、 Pattern recognition
摘要: Proposes a genetic algorithm-based approach to the design of compact fuzzy rule-based classification systems. In our approach, IF-THEN rule is generated by assigning circular cone-type membership function and certainty grade each training pattern. Thus, can be viewed as kind nearest-neighbor classifier, which has its own well localized receptive field specified radius function. A algorithm employed for selecting small number patterns that are used generating rules. Our three objectives: minimize error rate, rejection rate We also show system constructed represented neural network architecture similar networks.