Signal-based autonomous clustering for relational data

Relational datasets are defined by their relations, or their similarities, with each other. In this paper, we present a new autonomous clustering algorithm for relational data. We propose to use the properties of reordering methods in order to produce a one-dimensional signal of pairwise distances. A signal processing method is then applied to this signal to detect peaks of distance that reflect the clusters' separation. The main advantage of this procedure is that there is no parameter to tweak in order to obtain a reasonable clustering. The proposed method is compared with state-of-the-art algorithms to demonstrate it's efficiency.