As mentioned in Chapter 1, Getting Started with Machine Learning and ML.NET, anomaly detection, by definition, is an unsupervised learning algorithm. This means that the algorithm will train on data and look for data that does not fit the normal data. In this section, we will dive into use cases for anomaly detection and into the various trainers available for anomaly detection in ML.NET.
Use cases for anomaly detection
Anomaly detection, as you might have realized already, has numerous applications where data is available but it is unknown whether there is an anomaly in the data. Without needing to do manual spot-checking, anomaly detection algorithms train on this data and determine whether there are any anomalies. ML.NET provides various anomaly detection values to look at programmatically inside of your application. We will review these values later on in this chapter to better ensure that any detection is not a false positive.
Some of the potential...
