Tushych A. Methods of building an intelligent data analysis system based on neural networks

Tushych A.M. Methods of building an intelligent data analysis system based on neural networks. – Qualifying scientific work on the rights of the manuscript. Dissertation for the degree of Doctor of Philosophy in the specialty 123 Computer Engineering. – State University of Telecommunications of the Ministry of Education and Science of Ukraine, Kyiv, 2021. The dissertation is devoted to the development of methods for building an intelligent data analysis system based on neural networks. The paper formulates requirements for modern data analysis systems to justify the choice of the mathematical apparatus of the core of the analytical system. A method of nonlinear data normalization has been developed, which is based on the sequential execution of transformations of the variable type of nonlinearity, with the help of which a more uniform distribution can be achieved, which leads to increased learning efficiency of the neural network. The neural networks have an advantage over traditional methods of mathematical statistics and technical analysis for a given problem. The optimal topology having the structure of a multilayer perceptron and the optimal number of parameters of the selected topology are selected. A new approach has been developed, which consists in clustering the input parameters and activity of neurons, which allows to obtain the rules of data dependence on the trained neural network. The proposed algorithm allows you to detect hidden patterns of data in the form of rules that are clear to the user. The information technology of data mining with the help of the neural network device is developed, which allows to check the efficiency of the developed algorithms. The obtained results were used in research and economic contract works conducted at the State University of Telecommunications. Theoretical and practical provisions of the dissertation are used in the educational process of the State University of Telecommunications. Key words: information technologies, artificial intelligence, neural networks, clustering, data mining, knowledge discovery, big data.