Research on Intrusion Detection Model Based on DAE- XGBoost
Gao Minghui, Zhao Hang, Li Ma, Zhijun Zhang, Lu Kai, Cui Xudong, He Jicheng, Ning Zhiyan · 2022 IEEE 10th International Conference on Information, Communication and Networks (ICICN) · 2022
The DAE-XGBoost detection model designed in this paper solves the problem that the detection effect of rare attack types in massive network data is not ideal. First, extract information from large-scale high-dimensional data through DAE (Denoising Auto-encoder) neural network.Second, a trade-off factor is introduced into the XGBoost (eXtrame Gradient Boosting) model to balance the detection ability and improve the detection accuracy of rare attack types. The experiment uses the NSL-KDD data set for simulation verification. The results show that the detection accuracy of the proposed model for the two rare attack types “U2R” and “R2L” is 13.32% higher than the comparison method on average.