An Interpretable Deep Architecture for Similarity Learning Built Upon Hierarchical Concepts | IEEE Journals & Magazine | IEEE Xplore