Enhanced Lasso Recovery on Graph

Xavier Bresson; Thomas Laurent; James von Brecht

Enhanced Lasso Recovery on Graph

Xavier Bresson, Thomas Laurent, James von Brecht

Research output: Contribution to journal › Article › peer-review

Abstract

This work aims at recovering signals that are sparse on graphs. Compressed sensing offers techniques for signal recovery from a few linear measurements and graph Fourier analysis provides a signal representation on graph. In this paper, we leverage these two frameworks to introduce a new Lasso recovery algorithm on graphs. More precisely, we present a non-convex, non-smooth algorithm that outperforms the standard convex Lasso technique. We carry out numerical experiments on three benchmark graph datasets.

Original language	English
Pages (from-to)	1501-1505
Journal	Proceedings of the 23rd European Signal Processing Conference
State	Published - 2015

Keywords

Graph spectral analysis
Fourier basis
Lasso
l1 relaxation
sparse recovery
non-convex optimization

Disciplines

Mathematics

Access to Document

https://digitalcommons.lmu.edu/math_fac/92

Cite this

@article{40289eba9aeb47348c7bb0dde21c37a1,

title = "Enhanced Lasso Recovery on Graph",

abstract = " This work aims at recovering signals that are sparse on graphs. Compressed sensing offers techniques for signal recovery from a few linear measurements and graph Fourier analysis provides a signal representation on graph. In this paper, we leverage these two frameworks to introduce a new Lasso recovery algorithm on graphs. More precisely, we present a non-convex, non-smooth algorithm that outperforms the standard convex Lasso technique. We carry out numerical experiments on three benchmark graph datasets.",

keywords = "Graph spectral analysis, Fourier basis, Lasso, l1 relaxation, sparse recovery, non-convex optimization",

author = "Xavier Bresson and Thomas Laurent and {von Brecht}, James",

note = "X. Bresson, T. Laurent and J. von Brecht. {"}Enhanced Lasso Recovery.{"} Graph European Signal Processing Conference (EUSIPCO), 2015.",

year = "2015",

language = "English",

pages = "1501--1505",

journal = "Proceedings of the 23rd European Signal Processing Conference",

}

TY - JOUR

T1 - Enhanced Lasso Recovery on Graph

AU - Bresson, Xavier

AU - Laurent, Thomas

AU - von Brecht, James

N1 - X. Bresson, T. Laurent and J. von Brecht. "Enhanced Lasso Recovery." Graph European Signal Processing Conference (EUSIPCO), 2015.

PY - 2015

Y1 - 2015

N2 - This work aims at recovering signals that are sparse on graphs. Compressed sensing offers techniques for signal recovery from a few linear measurements and graph Fourier analysis provides a signal representation on graph. In this paper, we leverage these two frameworks to introduce a new Lasso recovery algorithm on graphs. More precisely, we present a non-convex, non-smooth algorithm that outperforms the standard convex Lasso technique. We carry out numerical experiments on three benchmark graph datasets.

AB - This work aims at recovering signals that are sparse on graphs. Compressed sensing offers techniques for signal recovery from a few linear measurements and graph Fourier analysis provides a signal representation on graph. In this paper, we leverage these two frameworks to introduce a new Lasso recovery algorithm on graphs. More precisely, we present a non-convex, non-smooth algorithm that outperforms the standard convex Lasso technique. We carry out numerical experiments on three benchmark graph datasets.

KW - Graph spectral analysis

KW - Fourier basis

KW - Lasso

KW - l1 relaxation

KW - sparse recovery

KW - non-convex optimization

M3 - Article

SP - 1501

EP - 1505

JO - Proceedings of the 23rd European Signal Processing Conference

JF - Proceedings of the 23rd European Signal Processing Conference

ER -