27
16

Optimal Analysis of Subset-Selection Based L_p Low Rank Approximation

Abstract

We study the low rank approximation problem of any given matrix AA over Rn×m\mathbb{R}^{n\times m} and Cn×m\mathbb{C}^{n\times m} in entry-wise p\ell_p loss, that is, finding a rank-kk matrix XX such that AXp\|A-X\|_p is minimized. Unlike the traditional 2\ell_2 setting, this particular variant is NP-Hard. We show that the algorithm of column subset selection, which was an algorithmic foundation of many existing algorithms, enjoys approximation ratio (k+1)1/p(k+1)^{1/p} for 1p21\le p\le 2 and (k+1)11/p(k+1)^{1-1/p} for p2p\ge 2. This improves upon the previous O(k+1)O(k+1) bound for p1p\ge 1 \cite{chierichetti2017algorithms}. We complement our analysis with lower bounds; these bounds match our upper bounds up to constant 11 when p2p\geq 2. At the core of our techniques is an application of \emph{Riesz-Thorin interpolation theorem} from harmonic analysis, which might be of independent interest to other algorithmic designs and analysis more broadly. As a consequence of our analysis, we provide better approximation guarantees for several other algorithms with various time complexity. For example, to make the algorithm of column subset selection computationally efficient, we analyze a polynomial time bi-criteria algorithm which selects O(klogm)O(k\log m) columns. We show that this algorithm has an approximation ratio of O((k+1)1/p)O((k+1)^{1/p}) for 1p21\le p\le 2 and O((k+1)11/p)O((k+1)^{1-1/p}) for p2p\ge 2. This improves over the best-known bound with an O(k+1)O(k+1) approximation ratio. Our bi-criteria algorithm also implies an exact-rank method in polynomial time with a slightly larger approximation ratio.

View on arXiv
Comments on this paper