Neuro-fuzzy model of complex objects approximation with discrete output

This paper solves the task of complex objects approximation with a discrete output based on information approach to modeling. We propose a model of fuzzy rules and the inference algorithm on the rules, and describe the neuro-fuzzy model for generation of a knowledge base. The approximation of known data sets and comparison of the results with those of other authors is performed. Examples of knowledge bases generation of the expert diagnostic systems in medicine, oil industry and information security show effectiveness of the proposed approach.