Axak N. Methods and models of distributed intelligent processing of big data in specialized computer systems

In the dissertation, the actual scientific and applied problem of developing, researching and improving methods and models that are the basis of distributed technologies for processing big data in specialized computer systems based on artificial neural networks using an agent-based approach and increasing their efficiency in terms of the quality of solutions and the timing of obtaining results with the maximum efficiency of the use of resources was solved. The analysis of methodological bases of a big data processing for problem-oriented services’ mass application was carried out. The methodological approach to building a service-oriented environment was developed. Proposed generalized organization model of problem-oriented calculations; general structure of the multi-agent system for emergency response; model of user service categorization; the model of the web resource’s personalization process and the object model of the documents of the problem-oriented web-service, which together constitute the methodological basis for the design of problem-oriented services. The principle of designing an adaptive and personalized service-oriented system was proposed. The Cloud-fog-dew architecture for personalized service-oriented systems was proposed. The expediency of designing a multi-agent system of operational response was shown. The MapReduce Hadoop model for distributed neural processing of big data was proposed. Methods for distributing work, co-opting and coordinating agent actions in an agent-oriented system were improved. The methodological basis for the processing of large data by neural networks was further developed. The methods of displaying multidimensional data into a small space were further developed. The methodology of distributed intelligent halftone image processing were further developed. An experimental system for solving the problem of diagnostics and remote monitoring was developed. The obtained practical results were confirmed by the acts of implementation and prove the correctness of the theoretical positions of the dissertation, the high efficiency of the developed models, methods.