A Hybrid and Improved Isolation Forest Algorithm for Anomaly Detection

Abstract

Anomalies defined as patterns or data points that do not conform to a well-defined notion of normal behavior. Anomaly detection is a significant research problem that caters to the interest of a large number of research scientists. It is a significant step in every useful data mining framework. Several techniques involving one or more of the following fields, namely statistical analysis, machine learning, soft computing, deep learning, and information theory, which have used for making better anomaly detection systems. Anomaly detection finds its applications in various fields such as detecting malicious behavior in online social media networks, detecting fraud in credit card transactions, fault detection systems. This paper presents a hybrid anomaly detection algorithm that outperforms the existing Isolation forest algorithm. A basic introduction of the existing algorithms given and then a comparative study performed between the existing algorithms and our hybrid algorithm.