Developing a victorious strategy to the second strong gravitational lensing data challenge
Autor(a) principal: | |
---|---|
Data de Publicação: | 2022 |
Outros Autores: | , , , , , , , , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UFRGS |
Texto Completo: | http://hdl.handle.net/10183/254587 |
Resumo: | Strong lensing is a powerful probe of the matter distribution in galaxies and clusters and a relevant tool for cosmography. Analyses of strong gravitational lenses with deep learning have become a popular approach due to these astronomical objects’ rarity and image complexity. Next-generation surveys will provide more opportunities to derive science from these objects and an increasing data volume to be analysed. However, finding strong lenses is challenging, as their number densities are orders of magnitude below those of galaxies. Therefore, specific strong lensing search algorithms are required to discover the highest number of systems possible with high purity and low false alarm rate. The need for better algorithms has prompted the development of an open community data science competition named strong gravitational lensing challenge (SGLC). This work presents the deep learning strategies and methodology used to design the highest scoring algorithm in the second SGLC (II SGLC). We discuss the approach used for this data set, the choice of a suitable architecture, particularly the use of a network with two branches to work with images in different resolutions, and its optimization. We also discuss the detectability limit, the lessons learned, and prospects for defining a tailor-made architecture in a survey in contrast to a general one. Finally, we release the models and discuss the best choice to easily adapt the model to a data set representing a survey with a different instrument. This work helps to take a step towards efficient, adaptable, and accurate analyses of strong lenses with deep learning frameworks. |
id |
UFRGS-2_df89ce37fc8ddb64a6421a169df32919 |
---|---|
oai_identifier_str |
oai:www.lume.ufrgs.br:10183/254587 |
network_acronym_str |
UFRGS-2 |
network_name_str |
Repositório Institucional da UFRGS |
repository_id_str |
|
spelling |
De Bom, Clécio RoqueFraga, Bernardo Machado de OliveiraDias, Luciana OliviaSchubert, PatrickValentin, Manuel BlancoFurlanetto, CristinaMakler, MartínTeles, K.Albuquerque, M.P.Metcalf, Robert Benton2023-02-10T04:56:18Z20220035-8711http://hdl.handle.net/10183/254587001155652Strong lensing is a powerful probe of the matter distribution in galaxies and clusters and a relevant tool for cosmography. Analyses of strong gravitational lenses with deep learning have become a popular approach due to these astronomical objects’ rarity and image complexity. Next-generation surveys will provide more opportunities to derive science from these objects and an increasing data volume to be analysed. However, finding strong lenses is challenging, as their number densities are orders of magnitude below those of galaxies. Therefore, specific strong lensing search algorithms are required to discover the highest number of systems possible with high purity and low false alarm rate. The need for better algorithms has prompted the development of an open community data science competition named strong gravitational lensing challenge (SGLC). This work presents the deep learning strategies and methodology used to design the highest scoring algorithm in the second SGLC (II SGLC). We discuss the approach used for this data set, the choice of a suitable architecture, particularly the use of a network with two branches to work with images in different resolutions, and its optimization. We also discuss the detectability limit, the lessons learned, and prospects for defining a tailor-made architecture in a survey in contrast to a general one. Finally, we release the models and discuss the best choice to easily adapt the model to a data set representing a survey with a different instrument. This work helps to take a step towards efficient, adaptable, and accurate analyses of strong lenses with deep learning frameworks.application/pdfengMonthly notices of the royal astronomical society. Oxford. Vol. 515, no. 4 (Oct. 2022), p. 5121–5134Lentes gravitacionaisProcessamento de imagensRedes neuraisGravitational lensing : StrongMethods : NumericalTechniques : Image processingDeveloping a victorious strategy to the second strong gravitational lensing data challengeEstrangeiroinfo:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/openAccessreponame:Repositório Institucional da UFRGSinstname:Universidade Federal do Rio Grande do Sul (UFRGS)instacron:UFRGSTEXT001155652.pdf.txt001155652.pdf.txtExtracted Texttext/plain71396http://www.lume.ufrgs.br/bitstream/10183/254587/2/001155652.pdf.txt0c10309f7c9a9f89b107fe2e9badd760MD52ORIGINAL001155652.pdfTexto completo (inglês)application/pdf3758516http://www.lume.ufrgs.br/bitstream/10183/254587/1/001155652.pdf9ce4cd0311c85c34e0d5a598d8324c27MD5110183/2545872023-06-17 03:37:27.307369oai:www.lume.ufrgs.br:10183/254587Repositório de PublicaçõesPUBhttps://lume.ufrgs.br/oai/requestopendoar:2023-06-17T06:37:27Repositório Institucional da UFRGS - Universidade Federal do Rio Grande do Sul (UFRGS)false |
dc.title.pt_BR.fl_str_mv |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
title |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
spellingShingle |
Developing a victorious strategy to the second strong gravitational lensing data challenge De Bom, Clécio Roque Lentes gravitacionais Processamento de imagens Redes neurais Gravitational lensing : Strong Methods : Numerical Techniques : Image processing |
title_short |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
title_full |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
title_fullStr |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
title_full_unstemmed |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
title_sort |
Developing a victorious strategy to the second strong gravitational lensing data challenge |
author |
De Bom, Clécio Roque |
author_facet |
De Bom, Clécio Roque Fraga, Bernardo Machado de Oliveira Dias, Luciana Olivia Schubert, Patrick Valentin, Manuel Blanco Furlanetto, Cristina Makler, Martín Teles, K. Albuquerque, M.P. Metcalf, Robert Benton |
author_role |
author |
author2 |
Fraga, Bernardo Machado de Oliveira Dias, Luciana Olivia Schubert, Patrick Valentin, Manuel Blanco Furlanetto, Cristina Makler, Martín Teles, K. Albuquerque, M.P. Metcalf, Robert Benton |
author2_role |
author author author author author author author author author |
dc.contributor.author.fl_str_mv |
De Bom, Clécio Roque Fraga, Bernardo Machado de Oliveira Dias, Luciana Olivia Schubert, Patrick Valentin, Manuel Blanco Furlanetto, Cristina Makler, Martín Teles, K. Albuquerque, M.P. Metcalf, Robert Benton |
dc.subject.por.fl_str_mv |
Lentes gravitacionais Processamento de imagens Redes neurais |
topic |
Lentes gravitacionais Processamento de imagens Redes neurais Gravitational lensing : Strong Methods : Numerical Techniques : Image processing |
dc.subject.eng.fl_str_mv |
Gravitational lensing : Strong Methods : Numerical Techniques : Image processing |
description |
Strong lensing is a powerful probe of the matter distribution in galaxies and clusters and a relevant tool for cosmography. Analyses of strong gravitational lenses with deep learning have become a popular approach due to these astronomical objects’ rarity and image complexity. Next-generation surveys will provide more opportunities to derive science from these objects and an increasing data volume to be analysed. However, finding strong lenses is challenging, as their number densities are orders of magnitude below those of galaxies. Therefore, specific strong lensing search algorithms are required to discover the highest number of systems possible with high purity and low false alarm rate. The need for better algorithms has prompted the development of an open community data science competition named strong gravitational lensing challenge (SGLC). This work presents the deep learning strategies and methodology used to design the highest scoring algorithm in the second SGLC (II SGLC). We discuss the approach used for this data set, the choice of a suitable architecture, particularly the use of a network with two branches to work with images in different resolutions, and its optimization. We also discuss the detectability limit, the lessons learned, and prospects for defining a tailor-made architecture in a survey in contrast to a general one. Finally, we release the models and discuss the best choice to easily adapt the model to a data set representing a survey with a different instrument. This work helps to take a step towards efficient, adaptable, and accurate analyses of strong lenses with deep learning frameworks. |
publishDate |
2022 |
dc.date.issued.fl_str_mv |
2022 |
dc.date.accessioned.fl_str_mv |
2023-02-10T04:56:18Z |
dc.type.driver.fl_str_mv |
Estrangeiro info:eu-repo/semantics/article |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
format |
article |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://hdl.handle.net/10183/254587 |
dc.identifier.issn.pt_BR.fl_str_mv |
0035-8711 |
dc.identifier.nrb.pt_BR.fl_str_mv |
001155652 |
identifier_str_mv |
0035-8711 001155652 |
url |
http://hdl.handle.net/10183/254587 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.ispartof.pt_BR.fl_str_mv |
Monthly notices of the royal astronomical society. Oxford. Vol. 515, no. 4 (Oct. 2022), p. 5121–5134 |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFRGS instname:Universidade Federal do Rio Grande do Sul (UFRGS) instacron:UFRGS |
instname_str |
Universidade Federal do Rio Grande do Sul (UFRGS) |
instacron_str |
UFRGS |
institution |
UFRGS |
reponame_str |
Repositório Institucional da UFRGS |
collection |
Repositório Institucional da UFRGS |
bitstream.url.fl_str_mv |
http://www.lume.ufrgs.br/bitstream/10183/254587/2/001155652.pdf.txt http://www.lume.ufrgs.br/bitstream/10183/254587/1/001155652.pdf |
bitstream.checksum.fl_str_mv |
0c10309f7c9a9f89b107fe2e9badd760 9ce4cd0311c85c34e0d5a598d8324c27 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFRGS - Universidade Federal do Rio Grande do Sul (UFRGS) |
repository.mail.fl_str_mv |
|
_version_ |
1801225080762531840 |