Using Machine Learning Techniques for Outlier Detection Application

The aim of designing models that have the ability for adapting to their surroundings and have capability for learning from past data has gained attention from numerous spheres, comprising computing engineering, artificial intelligence, statistics, and mathematics. Numerous machine learning methods have evolved from research in the field. Supervised, unsupervised and reinforcement techniques have been applied for a variety of scientific and industrial applications. This paper introduces a description of a scheme for outlier detection application that is based on Isolation Forest machine learning technique. The scheme is an unsupervised learning method. The scheme also uses Random Forest machine learning technique for the application. This method is a supervised learning method. The learning techniques are tested on a sample system. A comparison of Isolation Forest technique with other unsupervised learning algorithms is presented, along with its comparison to supervised Random Forest Classifier algorithm.