Intrusion Detection Method Based on Improved K-Means Algorithm

Abstract

Data mining technology has a good application in the field of intrusion detection. For the problem that K-Means algorithm is difficult to process high-dimensional data, local optimal solution, and cannot determine K value, this paper proposes an improved K-Means algorithm. Firstly, the PCA algorithm is used to reduce the dimension of the data set, and then the Outlier detection is used to eliminate the Outliers that have a great influence on the final clustering result. Then, the initial clustering center point is selected based on the distance to avoid the local optimal solution. Finally, K- is used. The Means algorithm performs clustering to obtain an intrusion cluster. The experimental results show that compared with the common data mining-based intrusion detection algorithm, the proposed method has a good performance in detection rate and false positive rate, and its performance has also improved.