Hyperbox model for fuzzy rule evaluation in neural networks

A model that is suggested for pattern classification by using fuzzy sets in a neural network is modified to include fuzzy rule evaluation. The proposed model is aimed to be used for medical diagnosis applications. In this paper, two variations of the original model are described. The drawbacks and advantages of both models are discussed along with the implementation results. We used the maximum hyperbox size parameter (θ) in the first model but not in the second one. The effects of θ and the defuzzification methods are also examined only for the first model. The related learning algorithms, which adjust the minimum and the maximum points for hyperboxes that represent the fuzzy ranges, are given with the necessary changes