Anomaly Detection

Anomalies happen all around us, every day, anywhere from large manufacturing machines to public roads to national defenses and beyond. Anomaly detection is the process of finding these differences and outliers in any given data. It works by using certain historical data, to compare normal patterns with new data to find unusual fluctuations and anomalies. Examples are vibrations, temperate changes and spikes in network traffic.

Anomaly detection helps us prevent, mitigate, or respond to these unusual activities. Any values that are outside the normal data parameters can be identified, flagged, and investigated as soon as possible. Without proper tools, the response can be too late, which means that the process has already moved to the stage of no return. As a result, there can be a broken machine, fraudulent activity, or too late response to defense breaches. Instead, anomaly detection helps to pre-position certain thresholds that can be used to recognize these anomalies and have warning systems in place to react to them as fast as possible.