Investigation of network infrastructure control parameters for effective intellectual analysis

One of the important problems faced by current intrusion detection systems, monitoring systems is the processing of large data sets that complicates the intellectual processing of data. To detect a variety of cases of threats and violations, the monitoring system should control a large number of parameters. Therefore, selection of characteristics is an important stage in the construction of algorithms for machine learning. This stage is necessary to get rid of noise attributes and due to this improve the quality and speed up the work of algorithms. The conducted experiments confirm that the algorithms for selecting attributes using Random Forest and the methods of filtration effectively manage their task.