一种基于特征工程和集成学习的网络入侵检测方法

With the increase in Internet popularity, cyber attacks are constantly proliferating, and their complexity and stealth seriously threaten network security. In this study, a network intrusion detection method based on feature engineering and ensemble learning is proposed. Through feature engineering methods such as data cleaning, data balancing and feature selection, this method solves the problems of category imbalance, data high dimension and feature redundancy in network traffic data set. At the same time, the improved ensemble learning method based on hyperparameter optimization is used to solve the problems of performance degradation and insufficient generalization ability of traditional machine learning algorithms when classifying various network threats. In this study, four common algorithms were selected on the UNSW-NB15 dataset and tested and compared with them using evaluation indicators. Experimental results show that the proposed model can improve the detection accuracy and outperform the comparison algorithm in all indicators.