Abstract:

Aiming at the problem of network intrusion feature optimization, we proposed a network intrusion detection model based on hybrid particle swarm optimization algorithm selecting features to improve the network intrusion detection rate. Firstly, we took the detection rate of network intrusion as
the objective function for feature selection, and the network state features as the constraint conditions to establish the corresponding mathematical model. Secondly, we used the hybrid particle swarm optimization algorithm to find the optimal feature subset. Finally, we took support vector machine as classifier to build intrusion detection model, and carried out the verification experiment by using KDD1999 data on MATLAB2012 platform. The results show that the model can efficiently query the optimal features subset, and intrusion detection rate and efficiency are better than the classical intrusion detection model.

Key words: internet, selection feature, intrusion detection model, classifier, hybrid particle swarm optimization algorithm