ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1904.09841
16
1

Simple Heuristics Yield Provable Algorithms for Masked Low-Rank Approximation

22 April 2019
Cameron Musco
Christopher Musco
David P. Woodruff
ArXivPDFHTML
Abstract

In masked low−rank approximationmasked\ low-rank\ approximationmasked low−rank approximation, one is given A∈Rn×nA \in \mathbb{R}^{n \times n}A∈Rn×n and binary mask matrix W∈{0,1}n×nW \in \{0,1\}^{n \times n}W∈{0,1}n×n. The goal is to find a rank-kkk matrix LLL for which: cost(L) = \sum_{i=1}^{n} \sum_{j = 1}^{n} W_{i,j} \cdot (A_{i,j} - L_{i,j} )^2 \leq OPT + \epsilon \|A\|_F^2 , where OPT=min⁡rank−k L^cost(L^)OPT = \min_{rank-k\ \hat{L}} cost(\hat L)OPT=minrank−k L^​cost(L^) and ϵ\epsilonϵ is a given error parameter. Depending on the choice of WWW, this problem captures factor analysis, low-rank plus diagonal decomposition, robust PCA, low-rank matrix completion, low-rank plus block matrix approximation, and many problems. Many of these problems are NP-hard, and while some algorithms with provable guarantees are known, they either 1) run in time nΩ(k2/ϵ)n^{\Omega(k^2/\epsilon)}nΩ(k2/ϵ) or 2) make strong assumptions, e.g., that AAA is incoherent or that WWW is random. In this work, we show that a common polynomial time heuristic, which simply sets AAA to 000 where WWW is 000, and then finds a standard low-rank approximation, yields bicriteria approximation guarantees for this problem. In particular, for rank k′>kk' > kk′>k depending on the public coin partition numberpublic\ coin\ partition\ numberpublic coin partition number of WWW, the heuristic outputs rank-k′k'k′ LLL with cost(L)≤OPT+ϵ∥A∥F2(L) \leq OPT + \epsilon \|A\|_F^2(L)≤OPT+ϵ∥A∥F2​. This partition number is in turn bounded by the randomized communication complexityrandomized\ communication\ complexityrandomized communication complexity of WWW, when interpreted as a two-player communication matrix. For many important examples of masked low-rank approximation, including all those listed above, this result yields bicriteria approximation guarantees with k′=k⋅poly(log⁡n/ϵ)k' = k \cdot poly(\log n/\epsilon)k′=k⋅poly(logn/ϵ). Further, we show that different models of communication yield algorithms for natural variants of masked low-rank approximation. For example, multi-player number-in-hand communication complexity connects to masked tensor decomposition and non-deterministic communication complexity to masked Boolean low-rank factorization.

View on arXiv
Comments on this paper