Intrusion detection algorithm based on improved principal component analysis

Abstract

Intrusion detection algorithm is a hot topic in the field of computer and network safety. The large load and high dimensionality of original network data collected by network traffic collector will reduce the detection efficiency and accuracy. Therefore, dimensionality reduction is necessary for original network data. Principal component analysis (PCA) is a common method for dimensionality reduction, but it only extracts principal components from the input data without considering the explanatory power of principal components on the output variables. To address this problem, we improve on the conventional PCA algorithm and apply this method to intrusion detection. Experiment shows that the improved PCA algorithm achieves better effect in dimensionality reduction and intrusion detection.

Keywords:

• Principal component analysis
• Intrusion detection
• Data mining