Anomaly Detection Based on Enhanced DBScan Algorithm

As the Internet becomes more pervasive, it is very important that the security mechanisms of a system are designed so as to prevent unauthorized access to system resources and data. The paper proposes a clustering algorithm that exploits enhanced DBScan algorithm in anomaly detection. The algorithm that can be used for mass data processing turns into the hot research point of anomaly detection, to form normal behavior profile on the audit records and adjust the profile timely as the program behavior changed. The experimental result shows that the anomaly detecting based on enhanced DBScan algorithm can a higher detection rate and a low rate of false positives of DARPA data sets.