High accuracy single-layer free-space diffractive neuromorphic classifiers for spatially incoherent light

Free-space all-optical diffractive systems have shown promise for neuromorphic classification of objects without converting light to the electronic domain. While the factors that govern these systems have been studied for coherent light, the fundamental properties for incoherent light have not been...

Full description

Saved in:
Bibliographic Details
Published inOptics express Vol. 30; no. 8; pp. 12510 - 12520
Main Authors Léonard, François, Fuller, Elliot J, Teeter, Corinne M, Vineyard, Craig M
Format Journal Article
LanguageEnglish
Published United States Optical Society of America 11.04.2022
Online AccessGet full text

Cover

Abstract Free-space all-optical diffractive systems have shown promise for neuromorphic classification of objects without converting light to the electronic domain. While the factors that govern these systems have been studied for coherent light, the fundamental properties for incoherent light have not been addressed, despite the importance for many applications. Here we use a co-design approach to show that optimized systems for spatially incoherent light can achieve performance on par with the best linear electronic classifiers even with a single layer containing few diffractive features. This performance is limited by the inherent linear nature of incoherent optical detection. We circumvent this limit by using a differential detection scheme that achieves greater than 94% classification accuracy on the MNIST dataset and greater than 85% classification accuracy for Fashion-MNIST, using a single layer metamaterial.
AbstractList Free-space all-optical diffractive systems have shown promise for neuromorphic classification of objects without converting light to the electronic domain. While the factors that govern these systems have been studied for coherent light, the fundamental properties for incoherent light have not been addressed, despite the importance for many applications. Here we use a co-design approach to show that optimized systems for spatially incoherent light can achieve performance on par with the best linear electronic classifiers even with a single layer containing few diffractive features. This performance is limited by the inherent linear nature of incoherent optical detection. We circumvent this limit by using a differential detection scheme that achieves greater than 94% classification accuracy on the MNIST dataset and greater than 85% classification accuracy for Fashion-MNIST, using a single layer metamaterial.
Author Vineyard, Craig M
Léonard, François
Teeter, Corinne M
Fuller, Elliot J
Author_xml – sequence: 1
  givenname: François
  surname: Léonard
  fullname: Léonard, François
– sequence: 2
  givenname: Elliot J
  surname: Fuller
  fullname: Fuller, Elliot J
– sequence: 3
  givenname: Corinne M
  surname: Teeter
  fullname: Teeter, Corinne M
– sequence: 4
  givenname: Craig M
  surname: Vineyard
  fullname: Vineyard, Craig M
BackLink https://www.ncbi.nlm.nih.gov/pubmed/35472885$$D View this record in MEDLINE/PubMed
https://www.osti.gov/biblio/1860267$$D View this record in Osti.gov
BookMark eNpN0E9rHCEYx3EpKc2f9pA3ECSn9jCpjjo6xxC2TSGwl_Ys7jOPWYOrG3UC--4zZZPSkx4-fB_4nZOTlBMScsnZDReD_L5e3UilGNMfyBlno-wkM_rkv_8pOa_1iTEu9ag_kVOhpO6NUWck3IfHLXUAc3FwoDWkx4hddAcs1BfEru4dIJ2C9wto4QVpwrnkXS77bQAK0dUafMBSqc-FLrwFF-OBhgR5iwVTo3G50T6Tj97Fil_e3gvy58fq991997D--evu9qEDwXXrhBHcMKe5kmr0E9PGwzQKJrwZwWjcgO9HvUGJ_TSMXPSDcqD8NGg5TRp6cUGuj91cW7AVQkPYQk4JoVluBtYPekFfj2hf8vOMtdldqIAxuoR5rnapDj1b6mqh344USq61oLf7EnauHCxn9u_8dr2yx_kXe_WWnTc7nP7J973FKydggmU
CitedBy_id crossref_primary_10_1063_5_0191977
crossref_primary_10_1002_aisy_202300536
crossref_primary_10_1109_JSTQE_2022_3207056
crossref_primary_10_1364_OE_523923
crossref_primary_10_1063_5_0201761
crossref_primary_10_1364_OE_462370
Cites_doi 10.1021/acsphotonics.1c00526
10.1038/s41598-018-30619-y
10.1364/PRJ.415964
10.48550/arXiv.1702.05659
10.1515/nanoph-2020-0291
10.1002/admt.201800100
10.1364/AO.421844
10.1364/OL.44.005186
10.1364/AO.58.003179
10.1103/PhysRevLett.123.023901
10.1364/PRJ.8.000046
10.1126/science.aat8084
10.1109/JSTQE.2019.2921376
10.1109/5.726791
10.1117/1.AP.1.4.046001
10.1038/s41566-021-00796-w
ContentType Journal Article
DBID NPM
AAYXX
CITATION
7X8
OTOTI
DOI 10.1364/OE.455007
DatabaseName PubMed
CrossRef
MEDLINE - Academic
OSTI.GOV
DatabaseTitle PubMed
CrossRef
MEDLINE - Academic
DatabaseTitleList CrossRef

PubMed
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: https://proxy.k.utb.cz/login?url=http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
DeliveryMethod fulltext_linktorsrc
Discipline Physics
EISSN 1094-4087
EndPage 12520
ExternalDocumentID 1860267
10_1364_OE_455007
35472885
Genre Journal Article
GroupedDBID ---
123
29N
2WC
8SL
AAWJZ
ACGFO
ADBBV
AENEX
AKGWG
ALMA_UNASSIGNED_HOLDINGS
ATHME
AYPRP
AZSQR
AZYMN
BAWUL
BCNDV
CS3
DIK
DSZJF
DU5
E3Z
EBS
F5P
GROUPED_DOAJ
GX1
KQ8
M~E
NPM
OFLFD
OK1
OPJBK
OPLUZ
P2P
RNS
ROL
ROP
ROS
TR2
TR6
XSB
AAYXX
CITATION
7X8
OTOTI
ID FETCH-LOGICAL-c317t-383180a715459fd078fcd9303f89c87ebcf297be4e2d6913265ac5fd674dd7c23
ISSN 1094-4087
IngestDate Fri May 19 00:40:09 EDT 2023
Fri Apr 12 04:41:02 EDT 2024
Thu Sep 12 20:42:29 EDT 2024
Thu May 23 23:35:26 EDT 2024
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 8
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c317t-383180a715459fd078fcd9303f89c87ebcf297be4e2d6913265ac5fd674dd7c23
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
USDOE
ORCID 0000-0002-4973-6743
0000000249736743
OpenAccessLink https://doi.org/10.1364/oe.455007
PMID 35472885
PQID 2656201325
PQPubID 23479
PageCount 11
ParticipantIDs osti_scitechconnect_1860267
proquest_miscellaneous_2656201325
crossref_primary_10_1364_OE_455007
pubmed_primary_35472885
PublicationCentury 2000
PublicationDate 2022-04-11
PublicationDateYYYYMMDD 2022-04-11
PublicationDate_xml – month: 04
  year: 2022
  text: 2022-04-11
  day: 11
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle Optics express
PublicationTitleAlternate Opt Express
PublicationYear 2022
Publisher Optical Society of America
Publisher_xml – name: Optical Society of America
References Garg (oe-30-8-12510-R18) 2018; 3
Léonard (oe-30-8-12510-R7) 2021; 8
Lin (oe-30-8-12510-R1) 2018; 361
Li (oe-30-8-12510-R6) 2019; 1
Burgos (oe-30-8-12510-R5) 2021; 60
Janocha (oe-30-8-12510-R13) 2016; 25
Wu (oe-30-8-12510-R3) 2020; 8
Mengu (oe-30-8-12510-R2) 2020; 26
Mengu (oe-30-8-12510-R17) 2020; 9
Zhou (oe-30-8-12510-R10) 2021; 15
Jiao (oe-30-8-12510-R12) 2019; 44
Lecun (oe-30-8-12510-R15) 1998; 86
Colburn (oe-30-8-12510-R4) 2019; 58
Yan (oe-30-8-12510-R8) 2019; 123
Ryou (oe-30-8-12510-R9) 2021; 9
Chang (oe-30-8-12510-R11) 2018; 8
References_xml – volume: 8
  start-page: 2103
  year: 2021
  ident: oe-30-8-12510-R7
  publication-title: ACS Photonics
  doi: 10.1021/acsphotonics.1c00526
  contributor:
    fullname: Léonard
– volume: 8
  start-page: 12324
  year: 2018
  ident: oe-30-8-12510-R11
  publication-title: Sci. Rep.
  doi: 10.1038/s41598-018-30619-y
  contributor:
    fullname: Chang
– volume: 9
  start-page: B128
  year: 2021
  ident: oe-30-8-12510-R9
  publication-title: Photonics Res.
  doi: 10.1364/PRJ.415964
  contributor:
    fullname: Ryou
– volume: 25
  start-page: 49
  year: 2016
  ident: oe-30-8-12510-R13
  publication-title: Schedae Inform.
  doi: 10.48550/arXiv.1702.05659
  contributor:
    fullname: Janocha
– volume: 9
  start-page: 4207
  year: 2020
  ident: oe-30-8-12510-R17
  publication-title: Nanophotonics
  doi: 10.1515/nanoph-2020-0291
  contributor:
    fullname: Mengu
– volume: 3
  start-page: 1800100
  year: 2018
  ident: oe-30-8-12510-R18
  publication-title: Adv. Mater. Technol.
  doi: 10.1002/admt.201800100
  contributor:
    fullname: Garg
– volume: 60
  start-page: 4356
  year: 2021
  ident: oe-30-8-12510-R5
  publication-title: Appl. Opt.
  doi: 10.1364/AO.421844
  contributor:
    fullname: Burgos
– volume: 44
  start-page: 5186
  year: 2019
  ident: oe-30-8-12510-R12
  publication-title: Opt. Lett.
  doi: 10.1364/OL.44.005186
  contributor:
    fullname: Jiao
– volume: 58
  start-page: 3179
  year: 2019
  ident: oe-30-8-12510-R4
  publication-title: Appl. Opt.
  doi: 10.1364/AO.58.003179
  contributor:
    fullname: Colburn
– volume: 123
  start-page: 023901
  year: 2019
  ident: oe-30-8-12510-R8
  publication-title: Phys. Rev. Lett.
  doi: 10.1103/PhysRevLett.123.023901
  contributor:
    fullname: Yan
– volume: 8
  start-page: 46
  year: 2020
  ident: oe-30-8-12510-R3
  publication-title: Photonics Res.
  doi: 10.1364/PRJ.8.000046
  contributor:
    fullname: Wu
– volume: 361
  start-page: 1004
  year: 2018
  ident: oe-30-8-12510-R1
  publication-title: Science
  doi: 10.1126/science.aat8084
  contributor:
    fullname: Lin
– volume: 26
  start-page: 1
  year: 2020
  ident: oe-30-8-12510-R2
  publication-title: IEEE J. Select. Topics Quantum Electron.
  doi: 10.1109/JSTQE.2019.2921376
  contributor:
    fullname: Mengu
– volume: 86
  start-page: 2278
  year: 1998
  ident: oe-30-8-12510-R15
  publication-title: Proc. IEEE
  doi: 10.1109/5.726791
  contributor:
    fullname: Lecun
– volume: 1
  start-page: 1
  year: 2019
  ident: oe-30-8-12510-R6
  publication-title: Adv. Photonics
  doi: 10.1117/1.AP.1.4.046001
  contributor:
    fullname: Li
– volume: 15
  start-page: 367
  year: 2021
  ident: oe-30-8-12510-R10
  publication-title: Nat. Photonics
  doi: 10.1038/s41566-021-00796-w
  contributor:
    fullname: Zhou
SSID ssj0014797
Score 2.4673588
Snippet Free-space all-optical diffractive systems have shown promise for neuromorphic classification of objects without converting light to the electronic domain....
SourceID osti
proquest
crossref
pubmed
SourceType Open Access Repository
Aggregation Database
Index Database
StartPage 12510
Title High accuracy single-layer free-space diffractive neuromorphic classifiers for spatially incoherent light
URI https://www.ncbi.nlm.nih.gov/pubmed/35472885
https://search.proquest.com/docview/2656201325
https://www.osti.gov/biblio/1860267
Volume 30
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV1Lj9MwELZKERIXxJuysDKIW5TSOE6cHFFVtFqp9LKL9hY5jgORQlO1CQIO_AR-MzN2XkCRFi5R5SiJ5fk6_sbzIuQVFg3TzOdunKW-y5lWLm5DLss5FsMDjqHwHHL9Ljy75OdXwdVk8mMUtdTU6Vx9O5pX8j9ShTGQK2bJ_oNk-5fCAPwG-cIVJAzXa8kYgzQcqVSzx6btaPWX2i0lsGgn32vtgraAvy32QDG5UJ-1Y8pXfqpgcQvlKGTORY7NsE204QGjq2VZYiqgqj5qU7ipROt9TGE3O1PZWX_Z9dEbGNBjPe7A622sPBJiMySqYkgxabrEQxsq4pzP-5MDrdsWIUsMCQTqu-7vvQcm_LV973Iviw_dvfa0AoSOZRG9kYIFcxJs1naT1UfGWq3cemss-qKRikVGtjiq_P2Qg8Q2qzlmai_EsMN1Xv3fNr4-HNF49UKebFaJffQGuclAcaHGXH9f9V4pLmyznm6-baUqePR1_9Vf-M20Aj39d9vFcJiLu-ROa3zQNxZJ98hEb--TWyYIWB0ekALxRDs80TGe6IAnOsITHeOJjvBEAU-0xxMd8EQNnh6Sy7eri-WZ27bicBUQzNr1I9D9CymQcMd5BrwyV1kM9CePYhUJnaqcxSLVXLMsjD2wCQKpgjwLBc8yoZj_iEy31VY_IdSDpwNPppjcyYUM0zhjeeQpCRTAg91lRl52C5jsbMWV5A8BzcgJLm0CNBFrHSsMClN14tmOajPyolvxBLQlusDkVlfNIYF5hQzdi_CZx1YU_Uf8gAsWRcHT60zghNwe4P2MTOt9o58DPa3TU3Osc2qQ8xPWiJFH
link.rule.ids 230,315,783,787,867,888,27936,27937
linkProvider ISSN International Centre
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=High+accuracy+single-layer+free-space+diffractive+neuromorphic+classifiers+for+spatially+incoherent+light&rft.jtitle=Optics+express&rft.au=L%C3%A9onard%2C+Fran%C3%A7ois&rft.au=Fuller%2C+Elliot+J.&rft.au=Teeter%2C+Corinne+M.&rft.au=Vineyard%2C+Craig+M.&rft.date=2022-04-11&rft.issn=1094-4087&rft.eissn=1094-4087&rft.volume=30&rft.issue=8&rft.spage=12510&rft_id=info:doi/10.1364%2FOE.455007&rft.externalDBID=n%2Fa&rft.externalDocID=10_1364_OE_455007
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1094-4087&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1094-4087&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1094-4087&client=summon