HIERARCHICAL NEURAL NETWORKS

The PNS module is discussed as the building block for the synthesis of parallel, self-organizing, hierarchical, neural networks (PSHNN). The PNS consists of a prerejector (P-unit), a neural network (N-unit) and a statistical analysis unit (S-unit). The last two units together are also referred to as the NS unit. The P- and NS-units are fractile in nature, meaning that each such unit may itself consist of a number of parallel PNS modules. Through a mechanism of statistical acceptance or rejection of input vectors for classification, the sample space is divided into a number of subspaces. The input vectors belonging to each subspace are classified by a dedicated set of PNS modules. This strategy results in considerably higher accuracy of classification and better generalization as compared to previous neural network models. If the delta rule network is used to generate the N-unit, each subspace approximates a linearly separable space. In this sense, the total system becomes similar to a piecewise linear model. 1