Detecting anomalies in security events based on statistical analysis and large language models

Information security application
Authors:
Nemchinov A. V.Ovasapyan T. D.Zhukovskii E. V.
Abstract:

The possibilities of using large language models and statistical methods to automate the detection of anomalies in OS security events are investigated. A method for detecting anomalies is proposed that allows to automatically identify significant deviations and form their interpretation. A software prototype implementing this method has been developed and tested.