A fuzzy neural network with trapezoid fuzzy weights
Hisao Ishibuchi, Kouichi Morioka, Hideo Tanaka · 1994
Proposes a fuzzy neural network architecture whose weights are given as trapezoid fuzzy numbers. The proposed fuzzy neural network can handle fuzzy inputs as well as real inputs. In both cases, outputs from the fuzzy neural network are fuzzy numbers. Next, we derive a learning algorithm from a cost function defined for level sets (i.e. /spl alpha/-cuts) of fuzzy outputs and fuzzy targets. Lastly, we examine the ability of the proposed fuzzy neural network to implement fuzzy IF-THEN rules by computer simulation.>