An approach for generation of perception frames based fuzzy neural network from data
Aleksandar Milosavljević, Leonid V. Stoimenov, Dejan Rančić · WSEAS TRANSACTIONS on SYSTEMS archive · 2008
In this paper we present an approach for generation and initialization of fuzzy neural networks (FNN) from data. Fuzzy neural networks are concept that integrates some features of the fuzzy logic and the artificial neural networks theory. Based on analysis of several different fuzzy neural networks models, uniform representation method is presented, and two basic types are identified: FNN based on perception frames, and FNN with independent rules. Presented algorithm supports generation of fuzzy neural network based on perception frames through following steps: clustering of a training set, identification of input variables perception frames, generation of rules using training set, and training for adaptation using gradient descent method.