At this stage, the high dimension and large variety of network data have increased the difficulty of intrusion detection. In this paper, we discuss the advantages and disadvantages of the MDBoost algorithm. Subsequently to optimize it, we add a slack variable in the objective function, so that the algorithm can effectively prevent over fitting, and the accuracy of the prediction is also improved. Then, we propose a model, which uses the MDBoost-2 algorithm to generate a strong classifier, and we use this model for intrusion detection. Finally, we use the CUP KDD 1999 data set to carry out the experiment. The results show that the new approach outperforms MDBoost and other well-known methods.

In the research of intrusion detection, we mainly focus on how to improve the accuracy of detection. Based on the introduction of support vector machine, this paper proposes a cubic polynomial smooth support vector machine model and uses it into intrusion detection. Subsequently we analyze each part of the model. Finally we conducted experiments to show that the proposed algorithm has higher accuracy than similar algorithms in Intrusion Detection.