Application of Artificial Neural Networks to Reveal Abnormal Behavior of Data Center Users

The purpose of the article: development of the method for detecting anomalous behavior of users of data centers based on the use of artificial neural networks. Research method: theoretical and system analysis of open data sources for detecting SQL queries and creating artificial neural networks, development and software implementation of a method for detecting anomalous behavior of data center users using artificial neural networks, experimental evaluation of the developed method. The result obtained: an approach to detecting anomalous behavior of users of data centers is proposed, based on the introduction of an analytical block containing a module of artificial neural networks into the protection system. The structure of an artificial neural network is proposed in the form of seven sequentially connected neural layers of a fixed dimension with different activation functions. The procedure for generating a data set for training a neural network based on a set of database log records is described. Examples of the implementation and experimental evaluation of the proposed method are given, confirming its effectiveness and high efficiency. The area of use of the proposed approach is anomaly and cyberattack detection components designed to improve the efficiency of information security monitoring and management systems.