Application for Traffic Classification Using Machine Learning Algorithms

The timely detection of malicious traffic and the prevention of DDoS attacks is an important task in today's informational environment. There was considered a theoretical part of the classical machine learning algorithms (the principles of their work and the information about parameters). The Functional and non-functional requirements were identified then the action schemes and the use cases were presented. There was developed a traffic classification system appearing as a window application. There was presented the implementation of the classical machine learning algorithms and a user's interface. The developed network traffic identification and classification system uses the machine learning algorithms: k-NN, Naive Bayes, SVM, Ridge / Lasso, Decision Tree and k-Means. The accuracy of the system ranged from 46.69% to 99.81% depending on which algorithm was used.