Multiview Clustering: A Scalable and Parameter-Free Bipartite Graph Fusion Method

Multiviewclustering partitions data into different groups according to their heterogeneous features. Most existingmethods degenerate the applicability ofmodels due to their intractable hyper-parameters triggered by various regularization terms. Moreover, traditional spectral basedmethods always encounter the expensive time overheads and fail in exploring the explicit clusters fromgraphs. In this paper, we present a scalable and parameter-free graph fusion framework for multiviewclustering, seeking for a joint graph compatible across multiple views in a self-supervisedweightingmanner. Our formulation coalescesmultiple view-wise graphs straightforward and learns theweights aswell as the joint graph interactively, which could actively release the model fromany weight-related hyper-parameters. Meanwhile, wemanipulate the joint graph by a connectivity constraint such that the connected components indicate clusters directly. The designed algorithmis initialization-independent and time-economicalwhich obtains the stable performance and scaleswell with the data size. Substantial experiments on toy data aswell as real datasets are conducted that verify the superiority of the proposedmethod compared to the state-of-the-arts over the clustering performance and time expenditure.