Hyperparameter-free sparse regression of grouped variables

Ted Kronvall; Stefan Ingi Adalbjornsson; Santhosh Nadig; Andreas Jakobsson

doi:10.1109/ACSSC.2016.7869067

Hyperparameter-free sparse regression of grouped variables

Ted Kronvall, Stefan Ingi Adalbjornsson, Santhosh Nadig, Andreas Jakobsson

Research output: Chapter in Book/Report/Conference proceeding › Paper in conference proceeding › peer-review

Abstract

In this paper, we introduce a novel framework for semi-parametric estimation of an unknown number of signals, each parametrized by a group of components. Via a reformulation of the covariance fitting criteria, we formulate a convex optimization problem over a grid of candidate representations, promoting solutions with only a few active groups. Utilizing the covariance fitting allows for a hyperparameter-free estimation procedure, highly robust against coherency between candidates, while still allowing for a computationally efficient implementation. Numerical simulations illustrate how the proposed method offers a performance similar to the group-LASSO for incoherent dictionaries, and superior performance for coherent dictionaries.

Original language	English
Title of host publication	Conference Record of the 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016
Publisher	IEEE Computer Society
Pages	394-398
Number of pages	5
ISBN (Electronic)	9781538639542
DOIs	https://doi.org/10.1109/ACSSC.2016.7869067
Publication status	Published - 2017 Mar 1
Event	50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016 - Pacific Grove, United States Duration: 2016 Nov 6 → 2016 Nov 9

Conference

Conference	50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016
Country/Territory	United States
City	Pacific Grove
Period	2016/11/06 → 2016/11/09

Subject classification (UKÄ)

Probability Theory and Statistics
Signal Processing

Keywords

convex optimization
covariance fitting
group sparsity
multi-pitch estimation

Access to Document

10.1109/ACSSC.2016.7869067

Cite this

@inproceedings{d1c97ff7ea12476684fb72369c241aed,

title = "Hyperparameter-free sparse regression of grouped variables",

abstract = "In this paper, we introduce a novel framework for semi-parametric estimation of an unknown number of signals, each parametrized by a group of components. Via a reformulation of the covariance fitting criteria, we formulate a convex optimization problem over a grid of candidate representations, promoting solutions with only a few active groups. Utilizing the covariance fitting allows for a hyperparameter-free estimation procedure, highly robust against coherency between candidates, while still allowing for a computationally efficient implementation. Numerical simulations illustrate how the proposed method offers a performance similar to the group-LASSO for incoherent dictionaries, and superior performance for coherent dictionaries.",

keywords = "convex optimization, covariance fitting, group sparsity, multi-pitch estimation",

author = "Ted Kronvall and Adalbjornsson, {Stefan Ingi} and Santhosh Nadig and Andreas Jakobsson",

year = "2017",

month = mar,

day = "1",

doi = "10.1109/ACSSC.2016.7869067",

language = "English",

pages = "394--398",

booktitle = "Conference Record of the 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016",

publisher = "IEEE Computer Society",

address = "United States",

note = "50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016 ; Conference date: 06-11-2016 Through 09-11-2016",

}

Kronvall, T, Adalbjornsson, SI, Nadig, S & Jakobsson, A 2017, Hyperparameter-free sparse regression of grouped variables. in Conference Record of the 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016., 7869067, IEEE Computer Society, pp. 394-398, 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016, Pacific Grove, United States, 2016/11/06. https://doi.org/10.1109/ACSSC.2016.7869067

Hyperparameter-free sparse regression of grouped variables. / Kronvall, Ted; Adalbjornsson, Stefan Ingi; Nadig, Santhosh; Jakobsson, Andreas.

Conference Record of the 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016. IEEE Computer Society, 2017. p. 394-398 7869067.

Research output: Chapter in Book/Report/Conference proceeding › Paper in conference proceeding › peer-review

TY - GEN

T1 - Hyperparameter-free sparse regression of grouped variables

AU - Kronvall, Ted

AU - Adalbjornsson, Stefan Ingi

AU - Nadig, Santhosh

AU - Jakobsson, Andreas

PY - 2017/3/1

Y1 - 2017/3/1

N2 - In this paper, we introduce a novel framework for semi-parametric estimation of an unknown number of signals, each parametrized by a group of components. Via a reformulation of the covariance fitting criteria, we formulate a convex optimization problem over a grid of candidate representations, promoting solutions with only a few active groups. Utilizing the covariance fitting allows for a hyperparameter-free estimation procedure, highly robust against coherency between candidates, while still allowing for a computationally efficient implementation. Numerical simulations illustrate how the proposed method offers a performance similar to the group-LASSO for incoherent dictionaries, and superior performance for coherent dictionaries.

AB - In this paper, we introduce a novel framework for semi-parametric estimation of an unknown number of signals, each parametrized by a group of components. Via a reformulation of the covariance fitting criteria, we formulate a convex optimization problem over a grid of candidate representations, promoting solutions with only a few active groups. Utilizing the covariance fitting allows for a hyperparameter-free estimation procedure, highly robust against coherency between candidates, while still allowing for a computationally efficient implementation. Numerical simulations illustrate how the proposed method offers a performance similar to the group-LASSO for incoherent dictionaries, and superior performance for coherent dictionaries.

KW - convex optimization

KW - covariance fitting

KW - group sparsity

KW - multi-pitch estimation

UR - http://www.scopus.com/inward/record.url?scp=85016250766&partnerID=8YFLogxK

U2 - 10.1109/ACSSC.2016.7869067

DO - 10.1109/ACSSC.2016.7869067

M3 - Paper in conference proceeding

AN - SCOPUS:85016250766

SP - 394

EP - 398

BT - Conference Record of the 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016

PB - IEEE Computer Society

T2 - 50th Asilomar Conference on Signals, Systems and Computers, ACSSC 2016

Y2 - 6 November 2016 through 9 November 2016

ER -

Hyperparameter-free sparse regression of grouped variables

Abstract

Conference

Subject classification (UKÄ)

Keywords

Access to Document

Other files and links

Fingerprint

Cite this