Robust machine learning for computer vision in naval application
Autor(a) principal: | |
---|---|
Data de Publicação: | 2023 |
Tipo de documento: | Dissertação |
Idioma: | eng |
Título da fonte: | Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) |
Texto Completo: | https://www.repositorio.mar.mil.br/handle/ripcmb/846372 |
Resumo: | This thesis proposes the development of a resilient machine learning algorithm that can classify naval images for surveillance, search, and detection operations in vast coastal areas. However, real-world datasets may be affected by label noise introduced either through random inaccuracies or deliberate adversarial attacks, both of which can negatively impact the accuracy of machine learning models. Our innovative approach employs Rockafellian Risk Minimization (RRM) to combat label noise contamination. Unlike existing methodologies reliant on extensively cleaned datasets, our two-step process involves adjusting neural network weights and manipulating data point nominal probabilities to isolate potential data corruption effectively. This technique reduces the dependency on meticulous data cleaning, thereby promoting more efficient and timeeffective data processing. To validate the efficacy and reliability of the proposed model, we apply RRM in several parameter configurations to naval environment datasets and assess its classification accuracy against traditional methods. By leveraging the proposed model, we aim to bolster the robustness of ship detection models, paving the way for a novel, reliable tool that could improve automated maritime surveillance systems. |
id |
MB_759be570780ba375cd0318ae26fad6f0 |
---|---|
oai_identifier_str |
oai:www.repositorio.mar.mil.br:ripcmb/846372 |
network_acronym_str |
MB |
network_name_str |
Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) |
repository_id_str |
|
spelling |
Rangel, Gabriel CustódioEckstrand, Eric C.2023-09-29T16:25:59Z2023-09-29T16:25:59Z2023https://www.repositorio.mar.mil.br/handle/ripcmb/846372This thesis proposes the development of a resilient machine learning algorithm that can classify naval images for surveillance, search, and detection operations in vast coastal areas. However, real-world datasets may be affected by label noise introduced either through random inaccuracies or deliberate adversarial attacks, both of which can negatively impact the accuracy of machine learning models. Our innovative approach employs Rockafellian Risk Minimization (RRM) to combat label noise contamination. Unlike existing methodologies reliant on extensively cleaned datasets, our two-step process involves adjusting neural network weights and manipulating data point nominal probabilities to isolate potential data corruption effectively. This technique reduces the dependency on meticulous data cleaning, thereby promoting more efficient and timeeffective data processing. To validate the efficacy and reliability of the proposed model, we apply RRM in several parameter configurations to naval environment datasets and assess its classification accuracy against traditional methods. By leveraging the proposed model, we aim to bolster the robustness of ship detection models, paving the way for a novel, reliable tool that could improve automated maritime surveillance systems.Naval Postgraduate SchoolEngenharia de produção aplicada à pesquisa operacional e gestão da inovaçãoMachine learningComputer visionNeural networksRobust machine learning for computer vision in naval applicationinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisinfo:eu-repo/semantics/openAccessengreponame:Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB)instname:Marinha do Brasil (MB)instacron:MBORIGINALDissertacao- GabrielCustodioRangel.pdfDissertacao- GabrielCustodioRangel.pdfapplication/pdf2619364https://www.repositorio.mar.mil.br/bitstream/ripcmb/846372/1/Dissertacao-%20GabrielCustodioRangel.pdf6346b6622755e28732ce799149968433MD51LICENSElicense.txtlicense.txttext/plain; charset=utf-83272https://www.repositorio.mar.mil.br/bitstream/ripcmb/846372/2/license.txt8ff7ce654d5215cee2106f3e3b7eb37fMD52ripcmb/8463722023-09-29 13:27:05.455oai:www.repositorio.mar.mil.br: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Repositório InstitucionalPUBhttps://www.repositorio.mar.mil.br/oai/requestdphdm.repositorio@marinha.mil.bropendoar:2023-09-29T16:27:05Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) - Marinha do Brasil (MB)false |
dc.title.pt_BR.fl_str_mv |
Robust machine learning for computer vision in naval application |
title |
Robust machine learning for computer vision in naval application |
spellingShingle |
Robust machine learning for computer vision in naval application Rangel, Gabriel Custódio Machine learning Computer vision Neural networks Engenharia de produção aplicada à pesquisa operacional e gestão da inovação |
title_short |
Robust machine learning for computer vision in naval application |
title_full |
Robust machine learning for computer vision in naval application |
title_fullStr |
Robust machine learning for computer vision in naval application |
title_full_unstemmed |
Robust machine learning for computer vision in naval application |
title_sort |
Robust machine learning for computer vision in naval application |
author |
Rangel, Gabriel Custódio |
author_facet |
Rangel, Gabriel Custódio |
author_role |
author |
dc.contributor.author.fl_str_mv |
Rangel, Gabriel Custódio |
dc.contributor.advisor1.fl_str_mv |
Eckstrand, Eric C. |
contributor_str_mv |
Eckstrand, Eric C. |
dc.subject.por.fl_str_mv |
Machine learning Computer vision Neural networks |
topic |
Machine learning Computer vision Neural networks Engenharia de produção aplicada à pesquisa operacional e gestão da inovação |
dc.subject.dgpm.pt_BR.fl_str_mv |
Engenharia de produção aplicada à pesquisa operacional e gestão da inovação |
description |
This thesis proposes the development of a resilient machine learning algorithm that can classify naval images for surveillance, search, and detection operations in vast coastal areas. However, real-world datasets may be affected by label noise introduced either through random inaccuracies or deliberate adversarial attacks, both of which can negatively impact the accuracy of machine learning models. Our innovative approach employs Rockafellian Risk Minimization (RRM) to combat label noise contamination. Unlike existing methodologies reliant on extensively cleaned datasets, our two-step process involves adjusting neural network weights and manipulating data point nominal probabilities to isolate potential data corruption effectively. This technique reduces the dependency on meticulous data cleaning, thereby promoting more efficient and timeeffective data processing. To validate the efficacy and reliability of the proposed model, we apply RRM in several parameter configurations to naval environment datasets and assess its classification accuracy against traditional methods. By leveraging the proposed model, we aim to bolster the robustness of ship detection models, paving the way for a novel, reliable tool that could improve automated maritime surveillance systems. |
publishDate |
2023 |
dc.date.accessioned.fl_str_mv |
2023-09-29T16:25:59Z |
dc.date.available.fl_str_mv |
2023-09-29T16:25:59Z |
dc.date.issued.fl_str_mv |
2023 |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/masterThesis |
format |
masterThesis |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
https://www.repositorio.mar.mil.br/handle/ripcmb/846372 |
url |
https://www.repositorio.mar.mil.br/handle/ripcmb/846372 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.publisher.none.fl_str_mv |
Naval Postgraduate School |
publisher.none.fl_str_mv |
Naval Postgraduate School |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) instname:Marinha do Brasil (MB) instacron:MB |
instname_str |
Marinha do Brasil (MB) |
instacron_str |
MB |
institution |
MB |
reponame_str |
Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) |
collection |
Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) |
bitstream.url.fl_str_mv |
https://www.repositorio.mar.mil.br/bitstream/ripcmb/846372/1/Dissertacao-%20GabrielCustodioRangel.pdf https://www.repositorio.mar.mil.br/bitstream/ripcmb/846372/2/license.txt |
bitstream.checksum.fl_str_mv |
6346b6622755e28732ce799149968433 8ff7ce654d5215cee2106f3e3b7eb37f |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da Produção Científica da Marinha do Brasil (RI-MB) - Marinha do Brasil (MB) |
repository.mail.fl_str_mv |
dphdm.repositorio@marinha.mil.br |
_version_ |
1798310234442170368 |