I. Introduction
Multiview data are widely available in the real world, where the same sample can be described by different modalities and perspectives [1], [2], [3], [4]. Unsupervised clustering of multiview data is an essential research content in machine learning community [5], [6]. The goal of multiview clustering is to partition samples into different clusters by exploiting complementary information across views and sample correlations within views [7], [8], [9].