Density-based semi-supervised classification for intrusion detection

Abstract

In order to improve the classification performance of intrusion detection problems with only a small number of labeled samples, semi-supervised learning is applied into the field of network intrusion. A semi-supervised classification method based on data density (SSC-density) is proposed to implement intrusion detection and solve network intrusion detection problem with fewer label samples. Firstly, the intrusion detection data is numerically numbered and normalized; secondly, the density of each sample is calculated, and the data samples are divided into security points, boundary points and noise points based on the density, so as to determine the spatial structure of the data; thirdly, different strategies are used for semi-supervised learning on different types of samples to mine the implicit information of unlabeled samples and expand the number of labeled samples. Specifically, deleting noise points, semi-supervised learning is firstly performed on the data set composed of security points, and then semi-supervised learning is performed on the data set composed of boundary points. Finally, the labeled samples are trained to generate the final classifier to realize the intrusion detection of network data. Experiments are carried out on KDD CUP 99 data set, the experimental result shows that the proposed algorithm has good classification performance.