Robust multi-view learning via half-quadratic minimization

Abstract

Although multi-view clustering is capable to use more information than single view clustering, existing multi-view clustering methods still have issues to be addressed, such as initialization sensitivity, the specification of the number of clusters, and the influence of outliers. In this paper, we propose a robust multi-view clustering method to address these issues. Specifically, we first propose a multi-view based sum-of-square error estimation to make the initialization easy and simple as well as use a sum-of-norm regularization to automatically learn the number of clusters according to data distribution. We further employ robust estimators constructed by the half-quadratic theory to avoid the influence of outliers for conducting robust estimations of both sum-of-square error and the number of clusters. Experimental results on both synthetic and real datasets demonstrate that our method outperforms the state-of-the-art methods.