On multistage fuzzy neural network modeling

In the past couple of years, there has been increasing interest in the fusi on of neural networks and fuzzy logic. Most of the existing fuzzy neural ne twork (FNN) models have been proposed to implement different types of singl e-stage fuzzy reasoning mechanisms and inevitably they suffer from the dime nsionality problem when dealing with complex real-world problems. To addres s the problem, FNN modeling based on multistage fuzzy reasoning (MSFR) is p ursued here and two hierarchical network models, namely incremental type an d aggregated type, are introduced. The new models railed multistage FNN (MS FNN) model a hierarchical fuzzy rule set that allows the consequence of a r ule passed to another as a fact through the intermediate variables, From th e stipulated input-output data pairs, they can generate an appropriate fuzz y rule set through structure and parameter learning procedures proposed in this paper. In addition, we have particularly addressed the input selection problem of these two types of multistage network models and proposed two e fficient methods for them. The effectiveness of the proposed MSFNN models i n handling high-dimensional problems is demonstrated through various numeri cal simulations.