I recently started to look into anomaly detection. The deep learning approaches are trained on "normal" classes to build a classifier that can detect outliers (anomalies). Are these approaches able to classify these anomalies and if so how can they do it, if they have never seen these before?
1 Answer
$\begingroup$
$\endgroup$
You can always use a clustering algorithm to group the anomalies into groups that share similar features. However you can't really call that classification if you don't have classes to assign.