Scalable Feature Matching Across Large Data Collections

This article is concerned with matching feature vectors in a one-to-one fashion across large collections of datasets. Formulating this task as multidimensional assignment problem decomposable costs (MDADC), we develop fast algorithms time complexity roughly linear the number n datasets and space small fraction data size. These remarkable properties hinge on using squared Euclidean distance dissimilarity function, which can reduce (n2) problems between pairs to enable calculating fly. To our knowledge, no other method applicable MDADC possesses these scaling low-storage necessary large-scale applications. In numerical experiments, novel outperform competing methods show excellent computational optimization performances. An application neuroimaging database presented. The are implemented R package matchFeat available at github.com/ddegras/matchFeat. Supplementary materials for online.