Hierarchical Neuro-Fuzzy BSP Model - HNFB

This paper presents a new hybrid neuro-fuzzy model which is capable of learning structure and parameters by means of recursive binary space partitioning BSP. Introduction Neuro-fuzzy systems (NFSs) [1] combine the learning ability of artificial neural nets (ANNs) with the linguistic interpretation capacity of fuzzy inference systems (FISs) [2]. This work makes use of BSP (Binary Space Partitioning) for the creation of a new neuro-fuzzy system that avoids the weak aspects of the traditional NFSs: the low number of inputs they work with and their low capacity to create a structure of their own. Neuro–Fuzzy Model HNFB An HNFB (Hierarchical Neuro-Fuzzy BSP) cell is a neuro-fuzzy mini-system that performs fuzzy binary partitioning in a specific space. An HNFB system is made up of interconnections of HNFB cells. Figure 1 illustrates a three-level HNFB system.