Multi-View Iterative Random Projections On Big Data Clustering

In the last few years, multi-view clustering has arisen in a large variety of applications. Properly, the clustering performance is greatly improved by exploiting the rich information among the different views. To this end, we propose a new method which achieve an efficient multi-view clustering of large-scale data. The key idea is to integrate simultaneously the random projection across multiple views in clustering process and applying K-means several times, by increasing the data dimension after each convergence of K-means. Extensive experiments are conducted on a high-dimensional data set to compare the proposed method with a number of mono-view and multi-view baselines methods. Empirical evaluations show the potential and the effectiveness of our method in terms of accuracy, purity and normalized mutual information compared with several improved methods proposed in the literature.