This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Highlights • We propose a novel method, Multilinear Side-Information based Discriminant Analysis (MSIDA), for dimensionality reduction and classification of tensor data, when the data full class label is missing. MSIDA projects the input face tensor into a new multilinear subspace in which the margin between samples belonging to different classes is enlarged while the margin within samples belonging to same classes is reduced. Additionally, MSIDA reduces the dimension of each tensor mode. • On the face description level, we propose a new representation based on high order tensors. This representation combines different local descriptors, extracted at different scales, providing better discrimination. The proposed tensor representation is regarded as a new way for fusing local descriptors. • We empirically evaluate the proposed approach for face based identity and kinship verification on four challenging face benchmarks (LFW, Cornell KinFace, UB KinFace and TSkinface). Comparison against the state-of the-art methods demonstrates the efficiency of our approach.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.