ПРИМЕНЕНИЕ МЕТОДОВ МАШИННОГО ОБУЧЕНИЯ ДЛЯ ФОРМИРОВАНИЯ СЦЕНАРИЕВ ПОВЕДЕНИЯ ВРЕДОНОСНЫХ ПРОГРАММ.

The paper is devoted to the problem of computer-aided forming of scenarios, with the latter describing the behavior of malicious software. The means of step by step problem solution using machine learning methods were discussed, and the choice of specific methods was substantiated. Presenting the possible scenarios in the form of Bayesian networks was proposed. Applicability of this approach was exemplified. Moreover, there were described the requirements to learning sample, which, if met at the data acquisition stage, would increase the accuracy of operation of machine learning methods. The experiments were conducted to demonstrate the improvement of the result quality with improvement of sample quality. [ABSTRACT FROM AUTHOR]