Information Granulation-Based Multi-layer Hybrid Fuzzy Neural Networks: Analysis and Design

Abstract

Abstract. In this study, a new architecture and comprehensive design methodology of genetically optimized Hybrid Fuzzy Neural Networks (gHFNN) are introduced and a series of numeric experiments are carried out. The gHFNN architecture results from a synergistic usage of the hybrid system generated by combining Fuzzy Neural Networks (FNN) with Polynomial Neural Networks (PNN). FNN contributes to the formation of the premise part of the overall network structure of the gHFNN. The consequence part of the gHFNN is designed using PNN. Introductory RemarksEfficient modeling techniques should allow for a selection of pertinent variables and a formation of highly representative datasets. The models should be able to take advantage of the existing domain knowledge and augment it by available numeric data to form a coherent data-knowledge modeling entity. The omnipresent modeling tendency is the one that exploits techniques of Computational Intelligence (CI) by embracing fuzzy modeling [1-6], neurocomputing [7], and genetic optimization [8].In this study, we develop a hybrid modeling architecture, called genetically optimized Hybrid Fuzzy Neural Networks (gHFNN). In a nutshell, gHFNN is composed of two main substructures driven to genetic optimization, namely a fuzzy set-based fuzzy neural network (FNN) and a polynomial neural network (PNN). From a standpoint of rule-based architectures, one can regard the FNN as an implementation of the antecedent part of the rules while the consequent is realized with the aid of a PNN. The role of the FNN is to interact with input data, granulate the corresponding input spaces. In the first case (Scheme I) we concentrate on the use of simplified fuzzy inference. In the second case (Scheme II), we take advantage of linear fuzzy inference. The role of the PNN is to carry out nonlinear transformation at the level of the fuzzy sets formed at the level of FNN. The PNN that exhibits a flexible and versatile structure [9] is constructed on a basis of Group Method of Data Handling (GMDH [10])