Biclustering with dominant sets

Abstract

Biclustering can be defined as the simultaneous clustering of rows and columns in a data matrix and it has been recently applied to many scientific scenarios such as bioinformatics, text analysis and computer vision to name a few. In this paper we propose a novel biclustering approach, that is based on the concept of dominant-set clustering and extends such algorithm to the biclustering problem. In more detail, we propose a novel encoding of the biclustering problem as a graph so to use the dominant set concept to analyse rows and columns simultaneously. Moreover, we extend the Dominant Set Biclustering approach to facilitate the insertion of prior knowledge that may be available on the domain. We evaluated the proposed approach on a synthetic benchmark and on two computer vision tasks: multiple structure recovery and region-based correspondence. The empirical evaluation shows that the method achieves promising results that are comparable to the state-of-the-art and that outperforms competitors in various cases.