Features of modeling the attacks on the machine learning model using Markov decision-making processes

Abstract: The issues that arise when determining the problem of modeling the impact on artificial intelligence models that are integrated into the network infrastructure are considered. Various research methods are presented and characterized, including the MITER methodology used in constructing the network impact vector. The features of using Markov decision-making processes in modeling attack influences are considered. Their significance for various procedures for determining vector parameters is considered. When constructing the modeling specifics, the au-thors take into account the features of determining the vulnera-bilities of artificial intelligence systems. The vector impact of major exploits of vulnerabilities is being studied.

Keywords: artificial intelligence, teaching method, policy, strat-egy, markov process modeling, vulnerability, network attack