Prototype selection for interpretable classification

Jacob Bien, Robert Tibshirani

Prototype methods seek a minimal subset of samples that can serve as a distillation or condensed view of a data set. As the size of modern data sets grows, being able to present a…