Martovitsky V. Models and method of detection of anomalies of functioning of computer systems on the basis of machine learning technology

The dissertation is devoted to the development of models and methods for detecting anomalies of the functioning of computer systems on the basis of machine learning technology. The scientific results are: 1) the first model of the classification of the state of the system, which allows to allocate a plurality of states depending on the functional tasks, delimit the processes of the target functioning of the system and interface processes of interaction with the network infrastructure and use them in the methods of intellectual analysis to identify the anomalies of the functioning of distributed systems. This model provides an opportunity to provide local control of parameters for each process and based on the generated vector to detect abnormal effects on the system as a whole; 2) the method of classification of the network status on the basis of statistical parameters for the detection of anomaly in the information structure of the computer system is improved; 3) received a further development of the multiagent model of the system of collecting and storing information based on agents. The proposed methods and means allow to increase the reliability of detection of anomalies of the functioning of distributed computer systems in the conditions of cybernetic influences of the external and internal environment by constructing models and methods for solving this problem on the basis of intelligent data analysis technologies.