Multi-view subspace clustering via partition fusion

Multi-view clustering is an important approach for analyzing multi-view data in unsupervised way. Among various methods, the subspace has gained increasing attention due to its encouraging performance. Essentially, it integrates information into graphs, which are then fed spectral algorithm final results. However, performance may degrade noises existing each individual view or inconsistencies between heterogeneous features. Orthogonal current work, we propose fuse a partition space, enhances robustness of clustering. Specifically, generate multiple partitions and integrate them find shared partition. The proposed model unifies graph learning, generation basic partitions, weight learning. These three components co-evolve towards better quality outputs. We have conducted comprehensive experiments on benchmark datasets our empirical results verify effectiveness approach.