Fuzzy Mapping NetWork Using Hierarchical Genetically Learning Rules
WuMeng, HeZhenya · 中国邮电高校学报：英文版 · 1994
This paper presents a novel architecture for the approximate reasoning-based Fuzzy Adap-tive Mapping Network(FAMN) .The FAMN includes two components :(a)Fuzzy Inference Net-work ,which is composed of a three-layer network according to the structure of employed rules;(b)Rules Learning Adapter ,which is used to adjush the membership function of rules with the serial ge-netic algorithm .In particular ,unlike other adaptive learning methods ,learning is achieved by incur-porating the idea of multiple resolution .The tuning is first implemented with few rules at the coarsest resolution fo input /output variables ,then with many rules at the higher resolution untis the training precise required is obtained .The 2-D sine function net is constructed as an illustrative example.The results have shown the proposed learning algorithm has better performance.