Scalable robust spectral ensemble clustering

Many ensemble clustering algorithms usually can work well on small-scale datasets, but the same expected results can not be achieved on large-scale datasets as well as time-consuming. Therefore, it is very important to implement an efficient clustering ensemble algorithm with high scalability to deal with these specific datasets. In this paper, we propose a scalable clustering approach based on the framework of the robust spectral ensemble clustering (RSEC), named as SRSEC to cluster the datasets of different sizes. A robust and denoising representation for the co-association matrix not only can be learned through a low-rank constraint in a unified optimization framework, but also a subspace selection on the co-association matrix can be constructed to do the robust spectral ensemble clustering. Experimental results show that our method has better clustering results in five real-world databases, especially in the large size of the databases.