English

Approximate Weighted $CR$ Coded Matrix Multiplication

Information Theory 2020-11-20 v1 Numerical Analysis math.IT Numerical Analysis

Abstract

One of the most common, but at the same time expensive operations in linear algebra, is multiplying two matrices AA and BB. With the rapid development of machine learning and increases in data volume, performing fast matrix intensive multiplications has become a major hurdle. Two different approaches to overcoming this issue are, 1) to approximate the product; and 2) to perform the multiplication distributively. A \textit{CRCR-multiplication} is an approximation where columns and rows of AA and BB are respectively sampled with replacement. In the distributed setting, multiple workers perform matrix multiplication subtasks in parallel. Some of the workers may be stragglers, meaning they do not complete their task in time. We present a novel \textit{approximate weighted CRCR coded matrix multiplication} scheme, that achieves improved performance for distributed matrix multiplication.

Keywords

Cite

@article{arxiv.2011.09709,
  title  = {Approximate Weighted $CR$ Coded Matrix Multiplication},
  author = {Neophytos Charalambides and Mert Pilanci and Alfred Hero},
  journal= {arXiv preprint arXiv:2011.09709},
  year   = {2020}
}

Comments

5 pages, 1 figure, conference