AUTOMATING SAFETY SYSTEM MODELLING IN ORGANIZATIONS USING NEURAL NETWORKS

The article addresses the pressing issue of network traffic analysis and integration of organizational security systems using neural networks. It provides statistics on types of cyber threats and concludes that various kinds of cyberattacks are becoming increasingly complex. The authors review modern approaches to detecting anomalies in traffic, preventing cyberattacks, and analysing surveillance camera data; give special attention to designing an information-analytical system capable of identifying threats in network traffic and analysing video footage to detect potential offenders. The authors conduct a comparative study of different neural network models, which concludes that the accuracy of processing network traffic and video stream data have improved. After the research, the authors select recurrent neural networks (RNN) as the primary model for training; next, discuss the analysis of recurrent neural networks (RNN) for handling temporal dependencies in network traffic and convolutional neural networks (CNN) for facial recognition from the organization database. The article presents the results of system testing, conducts computer simulations, and evaluates its effectiveness in detecting network threats and identifying suspicious objects in video data.