DDoS Attack Detection Method Based on Machine Learning

To reduce the impact of distributed denial of service (DDoS) attacks on the network, this paper studies the problem of detection of DDoS attacks in the network, aiming to identify the optimal detection model of DDoS attacks. To solve this problem, five machine learning algorithms, logistic regression, support vector machine (SVM), K-Nearest Neighbor (KNN), decision tree, and random forest, are used to identify attack traffic. The validity and reliability of the above models are tested by designing all feature experiments and some feature experiments. Experimental results show that the random forest algorithm performs the best in all feature experiments and some feature experiments, with an accuracy of 99.99% and 99.42% respectively. The F1-Score, response rate, and accuracy of all feature experiments are better than those of some selected feature experiments. Furthermore, it is verified that all feature experiments can be used to predict more accurately. Finally, the model is evaluated by using the confusion matrix and ROC curve to further verify the robustness of the model, thus improving the reliability of the study and enhancing the network security.