Variable projection without smoothness

Aleksandr Aravkin; Dmitriy Drusvyatskiy; Tristan van Leeuwen

doi:10.48550/arXiv.1601.05011

Variable projection without smoothness

Aleksandr Aravkin, Dmitriy Drusvyatskiy, Tristan van Leeuwen

Research output: Working paper › Academic

Abstract

Variable projection is a powerful technique in optimization. Over the last 30 years, it has been applied broadly, with empirical and theoretical results demonstrating both greater efficacy and greater stability than competing approaches. In this paper, we illustrate the technique on a large class of structured nonsmooth optimization problems, with numerical examples in sparse deconvolution and machine learning applications.

Original language	English
Publisher	arXiv
Pages	1-15
DOIs	https://doi.org/10.48550/arXiv.1601.05011
Publication status	Published - 19 Jan 2016

Bibliographical note

15 pages, 7 figures

Keywords

math.OC
stat.CO
stat.ML
65K05, 65K10, 86-08

Access to Document

10.48550/arXiv.1601.05011

Cite this

@techreport{6131366bd5b54732bfea69952f93d9ff,

title = "Variable projection without smoothness",

abstract = "Variable projection is a powerful technique in optimization. Over the last 30 years, it has been applied broadly, with empirical and theoretical results demonstrating both greater efficacy and greater stability than competing approaches. In this paper, we illustrate the technique on a large class of structured nonsmooth optimization problems, with numerical examples in sparse deconvolution and machine learning applications.",

keywords = "math.OC, stat.CO, stat.ML, 65K05, 65K10, 86-08",

author = "Aleksandr Aravkin and Dmitriy Drusvyatskiy and Leeuwen, {Tristan van}",

note = "15 pages, 7 figures",

year = "2016",

month = jan,

day = "19",

doi = "10.48550/arXiv.1601.05011",

language = "English",

pages = "1--15",

publisher = "arXiv",

type = "WorkingPaper",

institution = "arXiv",

}

TY - UNPB

T1 - Variable projection without smoothness

AU - Aravkin, Aleksandr

AU - Drusvyatskiy, Dmitriy

AU - Leeuwen, Tristan van

N1 - 15 pages, 7 figures

PY - 2016/1/19

Y1 - 2016/1/19

N2 - Variable projection is a powerful technique in optimization. Over the last 30 years, it has been applied broadly, with empirical and theoretical results demonstrating both greater efficacy and greater stability than competing approaches. In this paper, we illustrate the technique on a large class of structured nonsmooth optimization problems, with numerical examples in sparse deconvolution and machine learning applications.

AB - Variable projection is a powerful technique in optimization. Over the last 30 years, it has been applied broadly, with empirical and theoretical results demonstrating both greater efficacy and greater stability than competing approaches. In this paper, we illustrate the technique on a large class of structured nonsmooth optimization problems, with numerical examples in sparse deconvolution and machine learning applications.

KW - math.OC

KW - stat.CO

KW - stat.ML

KW - 65K05, 65K10, 86-08

U2 - 10.48550/arXiv.1601.05011

DO - 10.48550/arXiv.1601.05011

M3 - Working paper

SP - 1

EP - 15

BT - Variable projection without smoothness

PB - arXiv

ER -

Variable projection without smoothness

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this