Research on Data Mining Detection Algorithms for Abnormal Data

摘要

Adequate protection against malicious attacks is required to enhance the security of the Internet. This paper briefly introduced a data mining algorithm for network anomaly data detection, i.e., the K-means clustering algorithm. The detection performance of the K-means algorithm was improved by introducing the density-based spatial clustering of applications with noise (DBSCAN) algorithm and adjusting the number of clustering centers autonomously with standard deviation and cross-entropy. Simulation experiments compared the optimized K-means algorithm with support vector machine (SVM) and traditional Kmeans algorithms in MATLAB software. It was found that the optimized K-means algorithm had the best detection performance and the least detection time among the three abnormal data detection algorithms;the performance of the optimized K-means algorithm decreased as the proportion of abnormal data in the detected data increased, but it remained to be the best.

关键词