Attack detection models using machine learning methods

В1е article discusses some machine learning-based attack detection techniques that are used to identify anomalies in ready-made traffic sets. An analysis of the current situation in network security is provided. The authors propose a model for detecting attacks using machine learning methods and consider the issues of selecting data for training classifiers according to pre-generated criteria. Pre-processing of the selected data has been carried out. The article provides a classification of Brute Force and DDoS attacks. Training neural networks consists of minimizing loss functions by selecting optimal neuron weights during the execution of one or another optimization iterative algorithm. A conclusion is made about the model optimality based on the “decision tree” in the matter of classification based on logical regression.