Fusion Methods & Applications – A Survey

Abstract

Data fusion is the process in which integration of multiple data sources produce more consistent, efficient, and useful information than that provided by any of the individual data source. The fusion methods can be broadly classified with the help of class information. The fusion methods such as serial feature fusion, parallel feature fusion and canonical correlation analysis (CCA) does not contain any class information. That is, they are unsupervised fusion methods. The methods such as cluster CCA, Generalized Multiview Analysis, Linear Discriminant Analysis, Multiview Discriminant Analysis, Discriminative Multiple CCA, Locality Preserving CCA (LPCCA), BGLPCCA, MGLPCCA contain class information. Applications of each fusion methods are also taken into consideration.