Research on Intrusion Detection Technology Based on Ensemble Learning
Yue Zhao, Gang Gan · International Conference on Frontiers of Electronics, Information and Computation Technologies · 2021
With the rapid development of Internet technology, the network scale and data have become increasingly complex, and various forms of Internet attacks and destruction have appeared one after another. In order to effectively protect the security of computer network, people are paying more attention to and attaching more importance to intrusion detection technology. In view of the low detection rate and redundant features of intrusion detection caused by overfitting, this paper put forwards an ensemble model based on a combination of decision tree, random forest and gradient boosting decision tree. First, normalize the data in the intrusion detection dataset, and then perform correlation analysis on the redundant features in the dataset, and delete some highly correlated and repetitive features, finally use the ensemble learning method to perform normalization standard dataset for training. For performance evaluation, the UNSW_NB15 intrusion detection dataset is used. Three machine learning methods, like decision tree, random forest and gradient boosting decision tree, are utilized for comparison. The results indicate that our model with ensemble method has achieved better detection efficiency and accuracy.