There is renewed interest in Anomaly Detection today due to the expanding universe of sensors that monitor everything from your pulse and home temperature to large complex industrial machines, all of which generate time series data. For Industrial use cases, the interest is in preventive and predictive maintenance to cut down on maintenance costs and downtime of equipment. Examples include predicting failures of HVAC systems and elevators for property management and identifying potential signals of malfunction in machine engines. Many different statistical techniques exist for anomaly detection with some being more popular than others. In this presentation, we will take a close look at the types of anomalies that occur in time series and requirements for anomaly detection algorithms based on use cases. We will then look at some of the algorithms and results from our own internal work.
You are at :Home»Publications»Conference Proceedings»CloudXchange»The Machines are Talking: The Ride of Anomaly Detection