Exact continuous relaxations of ℓ0-regularized criteria with non-quadratic data terms

We consider the minimization of ℓ0-regularized criteria involving non-quadratic data terms such as the Kullback-Leibler divergence and the logistic regression, possibly combined with an ℓ2 regularization. We first prove the existence of global minimizers for such problems and characterize their loca...

Full description

Saved in:
Bibliographic Details
Published in:Journal of global optimization Vol. 93; no. 3; pp. 651 - 699
Main Authors: Essafri, Mhamed, Calatroni, Luca, Soubies, Emmanuel
Format: Journal Article
Language:English
Published: Dordrecht Springer Nature B.V 01.11.2025
Springer Verlag
Subjects:
ISSN:0925-5001, 1573-2916
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract We consider the minimization of ℓ0-regularized criteria involving non-quadratic data terms such as the Kullback-Leibler divergence and the logistic regression, possibly combined with an ℓ2 regularization. We first prove the existence of global minimizers for such problems and characterize their local minimizers. Then, we propose a new class of continuous relaxations of the ℓ0 pseudo-norm, termed as ℓ0 Bregman Relaxations (B-rex). They are defined in terms of suitable Bregman distances and lead to exact continuous relaxations of the original ℓ0-regularized problem in the sense that they do not alter its set of global minimizers and reduce its non-convexity by eliminating certain local minimizers. Both features make such relaxed problems more amenable to be solved by standard non-convex optimization algorithms. In this spirit, we consider the proximal gradient algorithm and provide explicit computation of proximal points for the B-rex penalty in several cases. Finally, we report a set of numerical results illustrating the geometrical behavior of the proposed B-rex penalty for different choices of the underlying Bregman distance, its relation with convex envelopes, as well as its exact relaxation properties in 1D/2D and higher dimensions.
AbstractList We consider the minimization of ℓ0-regularized criteria involving non-quadratic data terms such as the Kullback-Leibler divergence and the logistic regression, possibly combined with an ℓ2 regularization. We first prove the existence of global minimizers for such problems and characterize their local minimizers. Then, we propose a new class of continuous relaxations of the ℓ0 pseudo-norm, termed as ℓ0 Bregman Relaxations (B-rex). They are defined in terms of suitable Bregman distances and lead to exact continuous relaxations of the original ℓ0-regularized problem in the sense that they do not alter its set of global minimizers and reduce its non-convexity by eliminating certain local minimizers. Both features make such relaxed problems more amenable to be solved by standard non-convex optimization algorithms. In this spirit, we consider the proximal gradient algorithm and provide explicit computation of proximal points for the B-rex penalty in several cases. Finally, we report a set of numerical results illustrating the geometrical behavior of the proposed B-rex penalty for different choices of the underlying Bregman distance, its relation with convex envelopes, as well as its exact relaxation properties in 1D/2D and higher dimensions.
We propose a new class of exact continuous relaxations of l0-regularized criteria involving non-quadratic data terms such as the Kullback-Leibler divergence and the logistic regression, possibly combined with an l2 regularization. We first prove the existence of global minimizers for such problems and characterize their local minimizers.Then, we propose the l0 Bregman Relaxation (B-rex), a continuous approximation of the l0 pseudo-norm defined in terms of suitable Bregman distances, which leads to an exact continuous relaxations of the original l0-regularized problem in the sense that it does not alter its set of global minimizers and reduces the non-convexity by eliminating certain local minimizers. Both features make the relaxed problem more amenable to be solved by standard non-convex optimization algorithms. In this spirit, we consider the proximal gradient algorithm and provide explicit computation of proximal points for the B-rex penalty in several cases. Finally, we report a set of numerical results illustrating the geometrical behavior of the proposed B-rex penalty for different choices of the underlying Bregman distance, its relation with convex envelopes, as well as its exact relaxation properties in 1D/2D and higher dimensions.
Author Soubies, Emmanuel
Essafri, Mhamed
Calatroni, Luca
Author_xml – sequence: 1
  givenname: Mhamed
  surname: Essafri
  fullname: Essafri, Mhamed
– sequence: 2
  givenname: Luca
  surname: Calatroni
  fullname: Calatroni, Luca
– sequence: 3
  givenname: Emmanuel
  surname: Soubies
  fullname: Soubies, Emmanuel
BackLink https://hal.science/hal-04451909$$DView record in HAL
BookMark eNotjsFOwzAQRC1UJNrCD3CyxImDYdeOm_pYVYUiVeIC4hhtE5umSuPWcaD0zB_wh3wJqcpptDNvRzNgvdrXlrFrhDsESO8bhLEZC5BaAGqZisMZ66NOlZAGRz3WB9NFGgAv2KBp1gBgxlr22dtsT3nkua9jWbe-bXiwFe0plr5uuHf89_sHRLDvbUWhPNiC56GMNpTEP8u44t0OsWupCN1HzguKxLt001yyc0dVY6_-dcheH2Yv07lYPD8-TScLsZJoonCpRRg5XTjpELVOlCsS0xlmlObdnSoJzlor84J0vsRCqyVRYSjVDgiNGrLbU--Kqmwbyg2Fr8xTmc0ni-zoQZJoNGA-sGNvTuw2-F1rm5itfRvqbl6mZJocQYXqD7YYZbs
ContentType Journal Article
Copyright The Author’s 2025. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
licence_http://creativecommons.org/publicdomain/zero
Copyright_xml – notice: The Author’s 2025. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
– notice: licence_http://creativecommons.org/publicdomain/zero
DBID 7SC
8FD
JQ2
L7M
L~C
L~D
1XC
DOI 10.1007/s10898-025-01527-z
DatabaseName Computer and Information Systems Abstracts
Technology Research Database
ProQuest Computer Science Collection
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
Hyper Article en Ligne (HAL)
DatabaseTitle Computer and Information Systems Abstracts
Technology Research Database
Computer and Information Systems Abstracts – Academic
Advanced Technologies Database with Aerospace
ProQuest Computer Science Collection
Computer and Information Systems Abstracts Professional
DatabaseTitleList Computer and Information Systems Abstracts

DeliveryMethod fulltext_linktorsrc
Discipline Engineering
Sciences (General)
Mathematics
EISSN 1573-2916
EndPage 699
ExternalDocumentID oai:HAL:hal-04451909v1
GroupedDBID -~C
.4S
.86
.DC
.VR
06D
0R~
0VY
199
1N0
203
29K
2J2
2JN
2JY
2KG
2LR
2~H
30V
4.4
406
408
409
40D
40E
5GY
5VS
67Z
6NX
78A
7SC
7WY
8FD
8FL
8TC
8UJ
95-
95.
95~
96X
AABHQ
AACDK
AAHNG
AAIAL
AAJBT
AAJKR
AANZL
AAPKM
AARTL
AASML
AATNV
AATVU
AAUYE
AAWCG
AAYIU
AAYQN
ABAKF
ABBBX
ABBRH
ABBXA
ABDBE
ABDZT
ABECU
ABFSG
ABFTV
ABHLI
ABHQN
ABJNI
ABJOX
ABKCH
ABKTR
ABMNI
ABMQK
ABNWP
ABQBU
ABRTQ
ABSXP
ABTEG
ABTHY
ABTKH
ABTMW
ABWNU
ABXPI
ACAOD
ACDTI
ACGFS
ACGOD
ACHSB
ACHXU
ACIWK
ACKNC
ACMDZ
ACMLO
ACOKC
ACOMO
ACPIV
ACSTC
ACZOJ
ADHHG
ADHIR
ADIMF
ADKNI
ADKPE
ADRFC
ADTPH
ADURQ
ADYFF
ADZKW
AEFQL
AEGAL
AEGNC
AEJHL
AEJRE
AEMSY
AENEX
AEOHA
AEPYU
AETLH
AEVLU
AEXYK
AEZWR
AFBBN
AFDZB
AFHIU
AFLOW
AFOHR
AFQWF
AFWTZ
AFZKB
AGAYW
AGDGC
AGJBK
AGMZJ
AGQEE
AGQMX
AGRTI
AGWIL
AGWZB
AGYKE
AHAVH
AHBYD
AHKAY
AHPBZ
AHSBF
AHWEU
AHYZX
AIAKS
AIGIU
AIIXL
AILAN
AITGF
AIXLP
AJRNO
AJZVZ
ALMA_UNASSIGNED_HOLDINGS
ALWAN
AMKLP
AMXSW
AMYLF
AMYQR
AOCGG
ARAPS
ARMRJ
ASPBG
ATHPR
AVWKF
AXYYD
AYFIA
AYQZM
AZFZN
B-.
BA0
BAPOH
BENPR
BSONS
CS3
CSCUP
D-I
DDRTE
DL5
DNIVK
DPUIP
DU5
EBLON
EBS
EIOEI
ESBYG
FEDTE
FERAY
FFXSO
FIGPU
FNLPD
FRRFC
FWDCC
GGCAI
GGRSB
GJIRD
GNWQR
GQ7
GQ8
GXS
HCIFZ
HF~
HG5
HG6
HMJXF
HQYDN
HRMNR
HVGLF
HZ~
I09
IAO
IHE
IJ-
IKXTQ
ITM
IWAJR
IXC
IZIGR
IZQ
I~X
I~Z
J-C
J0Z
JBSCW
JCJTX
JQ2
JZLTJ
K60
K6~
KDC
KOV
L7M
LAK
LLZTM
L~C
L~D
M2O
MA-
N9A
NB0
NPVJJ
NQJWS
O93
O9G
O9I
O9J
OAM
P19
P2P
P9M
PF0
PT4
PT5
QOK
QOS
R89
R9I
RHV
RNS
ROL
RPX
RSV
S16
S1Z
S27
S3B
SAP
SBE
SDD
SDH
SDM
SHX
SISQX
SJYHP
SNE
SNPRN
SNX
SOHCF
SOJ
SPISZ
SRMVM
SSLCW
STPWE
SZN
T13
TN5
TSG
TSK
TSV
TUC
TUS
U2A
UG4
UOJIU
UTJUX
VC2
W23
W48
WK8
YLTOR
Z45
ZMTXR
~EX
-Y2
1SB
1XC
2.D
28-
2P1
2VQ
5QI
88I
8AO
8FE
8FG
8G5
AARHV
AAYTO
ABJCF
ABQSL
ABULA
ABUWG
ACBXY
ADHKG
AEBTG
AEFIE
AEKMD
AFEXP
AFFHD
AFGCZ
AFKRA
AGGDS
AGQPQ
AI.
AJBLW
AMVHM
ARCSS
AZQEC
BBWZM
BDATZ
BEZIV
BGLVJ
BGNMA
BPHCQ
CAG
CCPQU
COF
DWQXO
EDO
EJD
FINBP
FRNLG
FSGXE
GNUQQ
GUQSH
H13
ITC
K6V
K7-
KOW
L6V
M0C
M2P
M4Y
M7S
N2Q
NDZJH
NU0
O9-
OVD
P62
PHGZM
PHGZT
PQBIZ
PQBZA
PQGLB
PQQKQ
PROAC
PTHSS
Q2X
R4E
RNI
RZC
RZE
RZK
S26
S28
SCLPG
T16
TEORI
UZXMN
VFIZW
VH1
ZWQNP
ZY4
ID FETCH-LOGICAL-h219t-f7e106f5df2f115543fd496f5967c1557320feee2cda5cb1d53baad9a75f0a193
ISICitedReferencesCount 0
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001598957500001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 0925-5001
IngestDate Sun Nov 23 06:21:08 EST 2025
Mon Nov 24 04:10:54 EST 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 3
Language English
License licence_http://creativecommons.org/publicdomain/zero/: http://creativecommons.org/publicdomain/zero
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-h219t-f7e106f5df2f115543fd496f5967c1557320feee2cda5cb1d53baad9a75f0a193
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ORCID 0000-0003-0571-6983
0000-0003-3887-1859
PQID 3274445131
PQPubID 29930
PageCount 49
ParticipantIDs hal_primary_oai_HAL_hal_04451909v1
proquest_journals_3274445131
PublicationCentury 2000
PublicationDate 2025-11-01
PublicationDateYYYYMMDD 2025-11-01
PublicationDate_xml – month: 11
  year: 2025
  text: 2025-11-01
  day: 01
PublicationDecade 2020
PublicationPlace Dordrecht
PublicationPlace_xml – name: Dordrecht
PublicationTitle Journal of global optimization
PublicationYear 2025
Publisher Springer Nature B.V
Springer Verlag
Publisher_xml – name: Springer Nature B.V
– name: Springer Verlag
SSID ssj0009852
Score 2.4302945
Snippet We consider the minimization of ℓ0-regularized criteria involving non-quadratic data terms such as the Kullback-Leibler divergence and the logistic regression,...
We propose a new class of exact continuous relaxations of l0-regularized criteria involving non-quadratic data terms such as the Kullback-Leibler divergence...
SourceID hal
proquest
SourceType Open Access Repository
Aggregation Database
StartPage 651
SubjectTerms Algorithms
Approximation
Convex analysis
Convexity
Criteria
Fines & penalties
Mathematics
Optimization
Regression analysis
Regularization
Sparsity
Title Exact continuous relaxations of ℓ0-regularized criteria with non-quadratic data terms
URI https://www.proquest.com/docview/3274445131
https://hal.science/hal-04451909
Volume 93
WOSCitedRecordID wos001598957500001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVAVX
  databaseName: SpringerLINK Contemporary 1997-Present
  customDbUrl:
  eissn: 1573-2916
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0009852
  issn: 0925-5001
  databaseCode: RSV
  dateStart: 19970101
  isFulltext: true
  titleUrlDefault: https://link.springer.com/search?facet-content-type=%22Journal%22
  providerName: Springer Nature
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1Rb9MwELbK4AEeEBugDQayECAQsuQkTZ08VrCpiFFNdMDeIieO1Uo07ZKmqvbMP-An8M_4JdzZTtMVgeCBl6h1rEbNfbm72Pd9R8jTSCiIapyzlKcZ63pZziAopOAMVSZQkEppZZpNiOEwOj-PTzud7w0XZvlFFEW0WsXz_2pqGANjI3X2H8y9_lEYgM9gdDiC2eH4V4Y_WiHtEUvQJ0WNBa5IV1m1FW-mvKH7LA44K00j-nJyCVkneA-UbZZ2ZbaYFeyilqo0gq5YRvoKfXj1m1zWyYrMwP9MHbFznapXldSWzv5-LKeOS2V2PQwZqbD07LotGRrN6nTiaBLTqSxqV9Xv1ib80JH0ttcmsfAat0PW3Bm7AAnzQ-6m5879CuwsZ9mXjX-2HRQdDoMNZ9tzUrU2bvdso6VfQgJvKNIRkgl9LFUMfcEu2wDYbPoP-qPk9M1xcvJ2-O75_IJhazLcwnd9Wq6R674IY3SdH0afWmXnyLR1Wv8dx8xy_Myti0IGM8aC2624b5KZszvktrMc7Vv07JJOXuyRWxvalHtk13n9ir5w0uQv75LPBly0BRfdABedafrj67crsKINrCjCil6BFUVYUQOre-Tj8dHZ6wFzrTnYGELcgmmRe7ynQ6V97WFKGmjVjWEg7okMvovA5zrPcz9TMsxST8HzL6WKpQg1l_DScJ_swCXzfUIlV2EeQRaNnJIw0mkEr_xRV4d-6qdw5oA8gVuWzK34SoJy6IP-SYJjHNX1Yh4vvQNy2NzRxD2LVRKg-iXMCLwHfz79kNxs0XtIdhZlnT8iN7LlYlKVj425fwKaMoWg
linkProvider Springer Nature
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Exact+continuous+relaxations+of+%E2%84%930-regularized+criteria+with+non-quadratic+data+terms&rft.jtitle=Journal+of+global+optimization&rft.au=Essafri%2C+Mhamed&rft.au=Calatroni%2C+Luca&rft.au=Soubies%2C+Emmanuel&rft.date=2025-11-01&rft.pub=Springer+Nature+B.V&rft.issn=0925-5001&rft.eissn=1573-2916&rft.volume=93&rft.issue=3&rft.spage=651&rft.epage=699&rft_id=info:doi/10.1007%2Fs10898-025-01527-z&rft.externalDBID=HAS_PDF_LINK
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0925-5001&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0925-5001&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0925-5001&client=summon