A Rate-Distortion One-Class Model and its Applications to Clustering

We study the problem of one-class classification, in which we seek a rule to separate a coherent subset of instances similar to a few positive examples from a large pool of instances. We find that the