PROJECT TITLE :

Robust Rank-Constrained Sparse Learning: A Graph-Based Framework for Single View and Multiview Clustering

ABSTRACT:

Graph-based clustering is an approach that seeks to partition data in accordance with a similarity graph. This approach has demonstrated impressive performance across a wide variety of tasks. The quality of the similarity graph has a significant impact on the results of the clustering process; however, it is challenging to produce a high-quality similarity graph, particularly when the data include noises and outliers. In this article, we propose a solution to this issue in the form of a method called robust rank constrained sparse learning, or RRCSL. In order to learn the optimal graph in a way that is robust, the L2,1 norm has been incorporated into the objective function of the sparse representation. In order to keep the data structure intact, we first construct an initial graph and then search within that graph's immediate surroundings. The learned graph can be directly used as the cluster indicator, and the final results can be obtained without the need for any additional postprocessing if a rank constraint is included in the algorithm. The method that was suggested can not only be used for clustering based on a single view, but it can also be extended to clustering based on multiple views. The superiority and robustness of the proposed framework have been demonstrated by a large number of experiments on both synthetic and data sets taken from the real world.