We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Mathematics > Statistics Theory

Title: Learning tensors from partial binary measurements

Abstract: In this paper we generalize the 1-bit matrix completion problem to higher order tensors. We prove that when $r=O(1)$ a bounded rank-$r$, order-$d$ tensor $T$ in $\mathbb{R}^{N} \times \mathbb{R}^{N} \times \cdots \times \mathbb{R}^{N}$ can be estimated efficiently by only $m=O(Nd)$ binary measurements by regularizing its max-qnorm and M-norm as surrogates for its rank. We prove that similar to the matrix case, i.e., when $d=2$, the sample complexity of recovering a low-rank tensor from 1-bit measurements of a subset of its entries is the same as recovering it from unquantized measurements. Moreover, we show the advantage of using 1-bit tensor completion over matricization both theoretically and numerically. Specifically, we show how the 1-bit measurement model can be used for context-aware recommender systems.
Comments: 26 pages
Subjects: Statistics Theory (math.ST); Information Theory (cs.IT); Optimization and Control (math.OC)
MSC classes: 62B10, 94A17, 15A69, 62D05
ACM classes: H.3.3; I.2.6
DOI: 10.1109/TSP.2018.2879031
Cite as: arXiv:1804.00108 [math.ST]
  (or arXiv:1804.00108v1 [math.ST] for this version)

Submission history

From: Navid Ghadermarzy [view email]
[v1] Sat, 31 Mar 2018 02:31:15 GMT (53kb)

Link back to: arXiv, form interface, contact.