A theoretically sound upper bound on the triplet loss for improving the efficiency of deep distance metric learning

January 19th, 2021