Lytvynenko V. Methods and tools of hybrid artificial immune systems in problems of the intelligent data analysis

The thesis is devoted to the problems of development of hybrid artificial immune systems for solving the problems of forecasting and classification. In the work it is scientifically substantiated the problem of working out the methods and means of information technologies for analysis and synthesis of hybrid artificial immune systems. The methods are directed towards solving the problems of intelligent analysis of data of any nature. Here are analyzed the basic advantages and limitations of artificial immune systems. The perspective schemes of hybridization of artificial immune systems which allow to reach appreciable synergetic effect while solving the problems mentioned are offered. In the thesis the system approach to construction of adaptive hybrid and updating of known models and methods of synthesis of the artificial immune systems is developed. The new methodology is created for construction of modified and hybrid methods and analysis algorithms for diverse data. The formal problem-independent description of methods and computing procedures on the basis of principles of functioning of the immune system is proposed. The problems of forecasting, directed on solving, classification and optimization, offer tools for development the structure of the hybrid and combined artificial immune systems. The technology of structurally-parametrical synthesis RBF-, wavelet-, and fuzzy- neural networks for solving the problems of forecasting and classification is proposed. On the basis of the offered methods, techniques, models and computing procedures it was created original information analytical system for solving the problems of processing the large data arrays.