Fuzzy Membership Function Elicitation using Plausible Neural Network.
Kuo-Chen Li, Dar-Jen Chang · 2006
The elicitation method of the fuzzy membership function depends on the interpretation of the membership function. This paper applies a recently developed neural network framework, Plausible Neural Network (PNN), to generate fuzzy membership functions automatically with or without class labeling based on the similarity and likelihood measurement. The approach combining supervised and unsupervised learning provides a flexible method to generate fuzzy membership functions with semantic or statistical meanings. The results of simulation experiments show that PNN is capable of generating interesting and adaptive fuzzy membership functions.