Development of a solution for identifying network attacks based on adaptive neuro-fuzzy networks ANFIS

Within the framework of this work, the application of algorithms of adaptive neuro-fuzzy networks ANFIS based on various fuzzy rules, which allow identifying various network attacks, is considered. The implemented neuro-fuzzy networks by means of fuzzy transformations - algorithms of Sugeno-Takagi, Takagi-Sugeno-Kang and Wang-Mendel allow to classify suspicious network traffic. The obtained experimental results showed that the ANFIS network with Takagi-Sugeno-Kanga fuzzy inference is the most effective in terms of various measures of classification accuracy. At the same time, the analysis of the performance of algorithms for neuro-neural traffic classification showed that the presented methods required insignificant computational resources. The developed modules can be used to process data obtained from the security information and event management system.