English

Deterministic tensor completion with hypergraph expanders

Machine Learning 2021-10-11 v4 Machine Learning Optimization and Control Statistics Theory Statistics Theory

Abstract

We provide a novel analysis of low-rank tensor completion based on hypergraph expanders. As a proxy for rank, we minimize the max-quasinorm of the tensor, which generalizes the max-norm for matrices. Our analysis is deterministic and shows that the number of samples required to approximately recover an order-tt tensor with at most nn entries per dimension is linear in nn, under the assumption that the rank and order of the tensor are O(1)O(1). As steps in our proof, we find a new expander mixing lemma for a tt-partite, tt-uniform regular hypergraph model, and prove several new properties about tensor max-quasinorm. To the best of our knowledge, this is the first deterministic analysis of tensor completion. We develop a practical algorithm that solves a relaxed version of the max-quasinorm minimization problem, and we demonstrate its efficacy with numerical experiments.

Keywords

Cite

@article{arxiv.1910.10692,
  title  = {Deterministic tensor completion with hypergraph expanders},
  author = {Kameron Decker Harris and Yizhe Zhu},
  journal= {arXiv preprint arXiv:1910.10692},
  year   = {2021}
}

Comments

35 pages, 4 figures. To appear in SIAM Journal on Mathematics of Data Science