Design of a Fuzzy Classifier Network Based on Hierarchical Clustering

In this article the issue of data based modeling is dealt with the help a network of uniform multivariate fuzzy classifiers. Within this framework the innovation consists in the specification of a hierarchical design strategy for such a network. Concretely, the two network specifying factors, namely the layout of the network structure and the classifier nodes configuration, will be addressed by a hierarchical clustering and selection strategy. The resulting network will represent the model’s complexity in terms of interconnections between fuzzy classifier nodes whereas the model’s vagueness and imprecision is captured within each fuzzy classifier node. Throughout the paper the network design and operation is illustrated with the help of an example.