Investigation and identification of signs of hidden attacks in the enterprise for machine learning algorithms

Common attack styles imply human exploitation, namely cyber attacks, unskilled workers, staff negligence, unhealthy workplace environment. The spread of threats in enterprises often creates a human factor. If the technical device is a well-functioning and well-coordinated mechanism, which enables to measure the parameters of malfunctions and eliminate them using the diagnostic equipment, then a new system component is needed to investigate hidden attacks. Enterprises and industry as a whole need an intelligent system of protection and detection of hidden threats based on machine learning algorithms. To detect hidden threats, a set of measures is required to identify signs, analyze all components, predict and make recommendations, which is shown in the sections. The article discusses the problems of creating a knowledge base of historical vulnerability data at enterprises. A study of the oversaturation of diagnostic information with signs and retraining of the neural network was also conducted. Studies of data processing methods and their application in practice are shown. The study of the statistics of the detection of attacks and vulnerabilities at enterprises and the analysis of the human factor from a historical point of view is included in the structures of the manifestation of hidden threats. This is one of the main criteria for identifying vulnerabilities. All the methods considered, the results of which can be seen in the article, are a suitable layer for reorganizing data into knowledge and are applicable in the following studies. There are applied problems associated with the need to improve the analysis of internal and external parameters of the object of study in order to detect hidden threats.