Publication Type

Conference Proceeding Article

Version

acceptedVersion

Publication Date

8-2020

Abstract

This paper studies the problem of sparse residual regression, i.e., learning a linear model using a norm that favors solutions in which the residuals are sparsely distributed. This is a common problem in a wide range of computer vision applications where a linear system has a lot more equations than unknowns and we wish to find the maximum feasible set of equations by discarding unreliable ones. We show that one of the most popular solution methods, iteratively reweighted least squares (IRLS), can be significantly accelerated by the use of matrix sketching. We analyze the convergence behavior of the proposed method and show its efficiency on a range of computer vision applications. The source code for this project can be found at https://github.com/Diwata0909/Sketched IRLS.

Keywords

Sparse residual regression, L1 minimization, Randomized algorithm, Matrix sketching

Discipline

Artificial Intelligence and Robotics

Research Areas

Intelligent Systems and Optimization

Publication

ECCV 2020 16th European Conference on Computer Vision

First Page

609

Last Page

626

Identifier

doi.org/10.1007/978-3-030-58610-2_36

City or Country

UK

Share

COinS