Research on Intrusion Detection Model Using Ensemble learning Methods

In the past few years, for security issues keep increasing and changing explosively, Intrusion Detection System (IDS) have been asked for discovering these threats more accurately and more rapidly. Many intrusion detection models have been proposed to meet the requirements. However, for this unbalanced data sample KDDcup99, some classifiers may have a good effect on big sample classes like J48 and RandomForest, others are good at classifying on small sample, but few of them can achieve balance. In this paper, an intrusion detection solution based on ensemble learning is put forward. In our model, we employ Bayesian network and RandomTree as base classifiers along with meta learning algorithms RandomCommitte and vote. To evaluate the model's performance, the KDDcup99 dataset is introduced.