Adaptive Neuro-fuzzy Reasoning Based On Weighted Fuzzy Rules

The concept of adaptive neuro-fuzzy reasoning mechanism based on the weighted fuzzy production rules is given at first in this paper. Then five new fuzzy reasoning algorithms based on the weighted fuzzy production rules was presented. Each new reasoning algorithm is exactly mapped to a fuzzy neural network. The fuzzy neural network is trained by using the improved BP algorithms. The weights of the weighted fuzzy rules can be acquired by training the fuzzy neural network. Then we have five adaptive neuro-fuzzy reasoning algorithms: min-mean adaptive neuro-fuzzy reasoning, mean-mean adaptive neuro-fuzzy reasoning, mean-max adaptive neuro-fuzzy reasoning, mean-max-defuzzification adaptive neuro-fuzzy reasoning and mean-defuzzification adaptive neuro-fuzzy reasoning. The experiments show that the reasoning accuracy can be improved by using the adaptive neuro-fuzzy reasoning mechanism based on the weighted fuzzy production rules.