Towards a Zero-One Law for Entrywise Low Rank Approximation

11/04/2018
by   Zhao Song, et al.
0

There are a number of approximation algorithms for NP-hard versions of low rank approximation, such as finding a rank-k matrix B minimizing the sum of absolute values of differences to a given matrix A, _rank-k BA-B_1, or more generally finding a rank-k matrix B which minimizes the sum of p-th powers of absolute values of differences, _rank-k BA-B_p^p. Many of these algorithms are linear time columns subset selection algorithms, returning a subset of poly(k n) columns whose cost is no more than a poly(k) factor larger than the cost of the best rank-k matrix. The above error measures are special cases of the following general entrywise low rank approximation problem: given an arbitrary function g:R→R_≥0, find a rank-k matrix B which minimizes A-B_g=∑_i,jg(A_i,j-B_i,j). A natural question is which functions g admit efficient approximation algorithms? Indeed, this is a central question of recent work studying generalized low rank models. We give approximation algorithms for every function g which is approximately monotone and satisfies an approximate triangle inequality, and we show both of these conditions are necessary. Further, our algorithm is efficient if the function g admits an efficient approximate regression algorithm. Our algorithm handles functions which are not even scale-invariant, such as the Huber loss function. Our algorithms have poly(k)-approximation ratio in general. We further improve the approximation ratio to (1+ϵ) for ℓ_1 loss when the entries of the error matrix are i.i.d. random variables drawn from a distribution μ of which (1+γ) moment exists, for an arbitrary small constant γ>0. We also show our moment assumption is necessary.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
10/30/2017

Approximation Algorithms for ℓ_0-Low Rank Approximation

We study the ℓ_0-Low Rank Approximation Problem, where the goal is, give...
research
07/16/2021

Single Pass Entrywise-Transformed Low Rank Approximation

In applications such as natural language processing or computer vision, ...
research
07/16/2018

A PTAS for ℓ_p-Low Rank Approximation

A number of recent works have studied algorithms for entrywise ℓ_p-low r...
research
04/22/2019

Low-Rank Approximation from Communication Complexity

In low-rank approximation with missing entries, given A∈R^n× n and binar...
research
10/30/2019

Optimal Analysis of Subset-Selection Based L_p Low Rank Approximation

We study the low rank approximation problem of any given matrix A over R...
research
06/02/2023

Fast (1+ε)-Approximation Algorithms for Binary Matrix Factorization

We introduce efficient (1+ε)-approximation algorithms for the binary mat...
research
02/13/2020

Efficient Distance Approximation for Structured High-Dimensional Distributions via Learning

We design efficient distance approximation algorithms for several classe...

Please sign up or login with your details

Forgot password? Click here to reset