Design of Interval Type-2 FCM-Based Neural Networks
Wook-Dong Kim, Sung‐Kwun Oh, Kisung Seo · 2016
This paper is concerned with a design methodology of an Interval Type-2FCM-based fuzzy neural network classifier. The hidden layer of the proposed architecture is realized by interval type-2 FCM clustering to deal with uncertainties remaining in input space. This interval type-2 FCM clustering run using two values of the fuzzification coefficient resulting in interval type-2 membership functions and then the membership grades of IT2 FCM are used as the output of the hidden layer. Local LSE-based learning is applied to adjust the connection weights depicted as linear functions between the hidden layer and the output layer. The effectiveness of the proposed classifier is discussed and analyzed with the aid of a diversity of machine learning data sets.