Comparison and Application Research of Unsupervised Anomaly Detection Algorithms
Jincheng Li, Simin Fan, Xiang Li · 2024
The rapid development of the Internet not only provides convenience but also causes a variety of abnormal information. In general, the influence of abnormal information on different fields is far more important than the information it contains. Given this, this paper aims to study and analyze the characteristics of abnormal traffic by discussing the differences between outliers and normal points. To this end, based on the discussion of unsupervised anomaly detection algorithms involving statistics, distance, density, clustering, and tree, as well as the comparative study of each algorithm, Finally, experiments were conducted on some mainstream algorithms to demonstrate their accuracy in handling various types of information and situations. Meanwhile, this paper predicts the future research trend of unsupervised anomaly detection algorithms.