Best Rank-One Tensor Approximation and Parallel Update Algorithm for CPD

arXiv1709.08336MaRDI QIDQ6511816

Author name not available (Why is that?)

Abstract: A novel algorithm is proposed for CANDECOMP/PARAFAC tensor decomposition to exploit best rank-1 tensor approximation. Different from the existing algorithms, our algorithm updates rank-1 tensors simultaneously in parallel. In order to achieve this, we develop new all-at-once algorithms for best rank-1 tensor approximation based on the Levenberg-Marquardt method and the rotational update. We show that the LM algorithm has the same complexity of first-order optimisation algorithms, while the rotational method leads to solving the best rank-1 approximation of tensors of size $2imes2imescdotsimes2$. We derive a closed-form expression of the best rank-1 tensor of $2imes2imes2$ tensors and present an ALS algorithm which updates 3 component at a time for higher order tensors. The proposed algorithm is illustrated in decomposition of difficult tensors which are associated with multiplication of two matrices.

This page was built for publication: Best Rank-One Tensor Approximation and Parallel Update Algorithm for CPD

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6511816)