Intrusion Detection Model Based on Double Training Technique

In this paper, a network intrusion detection model was proposed based on double training technique to improve the frequency detection performance and to advance the detection ability for the low-frequency and high serious attacks. Firstly, the important features were extracted from whole dataset according to PCA. Then, a network intrusion detection model was constructed based on the classifier trained with double training technique. In experiments, the decision tree, naive Bayes and KNN algorithms were used respectively to construct the intrusion detection models based on double training technique. The experimental results show that the models can enhance the performance of the intrusion detection, especially for the low-frequency attacks.