“…In a distinct (but not unrelated) second line of research, parametric models consisting of higher order tensors made their way in inductive as well as transductive learning methods [44,39,51]. Within transductive techniques, in particular, tensor completion and tensor recovery emerged as a useful higher order generalization of their matrix counterpart [30,19,45,44,49,33,29,10,31]. By dealing with the estimation of tensors in the unifying framework of reproducing kernel Hilbert spaces (RKHSs), this paper positions itself within this second line of research.…”