Fuzzy Relation-based Polynomial Neural Networks Based on Information Granulation and Symbolic Gene Type Genetic Optimization

Abstract—In this paper, we introduce a new architecture of Information Granulation-based genetically optimized Fuzzy Relation-based Polynomial Neural Networks (IG_gFRPNN) that is based on a genetically optimized multilayer perceptions with fuzzy Relation-based polynomial neurons (FRPNs). The proposed IG_gFRPNN gives rise to a structurally optimized network and comes with a substantial level of flexibility in comparison to the one we encounter in conventional FPNN. In addition, the fuzzy rules used in the networks rely directly upon the notion of information granules defined over system's variables and formed through the process of information granulation. This granulation is realized with the aid of the C-Means clustering algorithm. Through the consecutive process of such structural and parametric optimization, a flexible topology of the fuzzy neural network becomes generated in a dynamic fashion. To evaluate the performance of the IG_gFRPNN, the model is experimented with using time series data (Mackey-Glass time series). I.