Multi view clustering with fuzzy and view weighting
Antim Yadav, Shadab Ali
Internet and its related applications like social networking, online shopping, virtual classrooms etc are growing and access to more and more people around the world at a fast pace. A lot of information is collected through clustering from this data. Ordinarily clustering performed over a single view of data is not sufficient to derive proper information. Recently, practice of combining different views of data through clustering has increased. Many research works prove that such method, called multi-view clustering, produces better clustering results. As the dimensionality of data increases, inclusion of all dimensions in clustering becomes time consuming. Moreover, the semantics of data. Inherently gives preference to some attributes of data objects over others. This leads to feature selection as an essential step before clustering is performed. When multiple views are involved, a ranking system among the views can also benefit by producing results oriented towards what analyst desires. Hence, this dissertation focuses on designing a multi-view clustering method which involves feature selection and view weighing. scheme.