A Novel Distributed Anomaly Detection Algorithm for Low-density Data

摘要

Anomaly detection becomes an interesting and significant problem in data mining area. It, in turn, generates many promising applications such as Intrusion Detection, Fraud Detection, Mobile cellular network fraud, and etc. Most of the existing anomaly detection methods are specifically modeled on certain tasks or datasets. And the density-based model has widely applied in anomaly detection. However, for many real-world applications, domain experts usually confront with large scale low-density data. This paper aims to solve the anomaly detection problem of large-scale low-density data. The Relative Local Density-based Outlier Factor (RLDOF) is a highly efficient and promising method for low-density data. While it is not easy to achieve considerable results concerning large-scale data. In this paper, we propose a distributed RLDOF model for low-density data on a large scale. The experimental results indicate that, compared with the existing methods, the model can effectively improve the anomaly detection performance of large-scale low-density data.