We present here an automated classification scheme which is particularly well suited to
scenarios where the parameters have significant uncertainties or are stochastic quantities.
To this end, the parameters are modeled with probability distributions in a metric space and
classification is conducted using the notion of nearest neighbors. The presented framework
is then applied to the classification of type I and type III edge-localized modes (ELMs) from a
set of carbon-wall plasmas at JET. This provides a fast, standardized classification of ELM …