In order to improve the clustering performance of traditional multi-view K-means algorithm in high-dimensional data, this paper proposed a robust large-scale multi-view data self-reduction-dimensional K-means algorithm RMSKMC. It realized the self-dimension reduction of high-dimensional data by finding the optimal subspace on a single view, and reconstructed the loss function by non-negative matrix factorization(NMF), so that different views shared the same cluster indicator matrix to realize multi-view information complementation, complete clustering of large-scale multi-view data. The experimental results show that the algorithm consumes less resources and performs more accurate clustering than other multi-view clustering algorithms on large-scale multi-view datasets.

[2020-10-05] Printed Article /doc/2019.06.0206/v2