Abstract: The efficient estimation of an approximate model order is very important for real applications with multi-dimensional low-rank data that may be corrupted by additive noise. In this paper, we ...