A neural network for fast inferencing on a fuzzy knowledge base

Abstract

The authors discuss a framework where fast inferences could be made from a fuzzy knowledge base of examples using neural networks. They sketch a scheme for defuzzifying the example base into a set of similarity vectors using a trait-adjusted similarity measure between a pair of fuzzy terms, and then training a neural network with these similarity vectors. A multilayer feedforward neural network with backpropagation learning rule was used.<>