Spectral clustering as mapping to a simplex

Spectral methods have been widely used to study the structural properties of unlabeled datasets. In this work we describe a clustering approach that exploits the structural properties in the configuration space of objects as well as the spectral sub-space, quite unlike earlier methods. We propose a spectral clustering approach, where we formalize the notion of clusters as vertices of a simplex in the spectral subspace. We define clustering as memberships of data points to vertices of this simplex. We empirically demonstrate that our method is comparable to the state-of-theart methods in a variety of domains and outperforms other generic clustering algorithms