Abnormal Traffic Prediction and Classification based on Information Big Data
Yuyang Tian · Highlights in Science Engineering and Technology · 2022
Intrusion Detection System (IDS) is a proactive security technique for detecting and alerting suspicious signals. However, the intrusion method developed as a fast and traditional method for detecting malicious traffic has a lot of shortcomings like low accuracy and low efficiency. To determine the different intrusion methods' features and promote the accuracy of malicious traffic detection, several Machine Learning models for classifying different intrusion methods such as KNN, Naive Bayes, SVM, LightGBM are compared. To further improve the accuracy of the model, ensemble models like Voting, Stacking for comparison are also introduced. Grid Search is used for the best parameters. The accuracy, precision, recall score and F1 score are used as metrics to evaluate the performances of different models. The experimental comparison and analysis show that the integrated learning algorithm based on Stacking has the highest accuracy for malicious traffic detection.