Katerinich L. Neural methods for constructing expert systems with associative models of knowledge representation

Dissertation is devoted to the development of neural network methods for constructing expert systems with associative models of knowledge representation. The formal presentation models of associative knowledge and the mechanism of inference on them. Formulated and solved the problem of classification for some well-known associations in terms of clarity. The approaches to the synthesis and optimization of neural network architecture with variable switching element. We study neural network model analysis of fuzzy data, which describes the fuzzy specifications (algorithms). Implemented an experimental version of the diagnostic expert system H-Homeopath (in environment Microsoft Visual Studio 2008 using DBMS Microsoft SQL Server 2008), which provides information support to processes of diagnosis and retrieval of data for reference.