It is a challenging task to integrate multi-view representations, each of which is of high dimension to improve the clustering performance. In this paper, we aim to improve the clustering performance of spectral clustering method when the manifold for high-dimensional data is not well defined in the multiple-view setting. We abstract the discriminative information on each view by spectral embedded clustering which performs well on high-dimensional data without a clear low-dimensional manifold structure. We bootstrap the clusterings of different views using discriminative information from one another. We derive a co-training algorithm to obtain a most informative clustering by iteratively modifying the affinity graph used for one view using the discriminative information from the other views. The approach is based on the assumption that the clustering from one view should agree with the clustering from another view. Comprehensive experiments on four real-world multiple-view high-dimensional datasets are presented to demonstrate the effectiveness of the proposed approach.
展开▼
