An Anomaly Detection Approach Based on Isolation Forest Algorithm for Streaming Data using Sliding Window

Anomalous behavior detection in many applications is becoming more and more important, such as computer security, sensor network and so on. However, the inherent characteristics of streaming data, such as generated quickly, data infinite, tremendous volume and the phenomenon of concept drift, imply that the anomaly detection in the streaming data is a challenge work. In this paper, using the frame of sliding windows and taking into account the concept drift phenomenon, a novel anomaly detection framework is presented and an adapted streaming data anomaly detection algorithm based on the iForest algorithm, namely iForestASD is proposed. The experiment results performed on four real-world datasets derived from the UCI repository demonstrate that the proposed algorithm can effective to detect anomalous instances for the streaming data.