G.G. Kulikov, V.V. Antonov, D.V. Antonov

Problems of creation of information systems with given characteristics is always belonged to the number of hard-to-resolve. The direct application of analytical methods requires a deep level of pre-processing. Application of neural network techniques involves, first of all, with the complexity of data warehousing, extraction of them mentioned statistics in explicit form, there is no strict theory at the choice of the type and architecture of the neural network, which leads to the need to apply a self-organizing algorithms, which also are quite slow. Up to the present time is the semantic gap between the rich notions about the subject area and the means that are used to express these ideas in the form of formal specifications. Discusses the possibility of construction of neural networks, the model is based on the idea of using fuzzy statistics to determine the parameters of the functions of the accessories that best fit some system of fuzzy inference.