Análise da técnica deep forest para o problema de aprendizado de ranqueamento
Autor(a) principal: | |
---|---|
Data de Publicação: | 2022 |
Tipo de documento: | Dissertação |
Idioma: | por |
Título da fonte: | Repositório Institucional da UFG |
Texto Completo: | http://repositorio.bc.ufg.br/tede/handle/tede/12087 |
Resumo: | Learning to Rank (LeToR) is a specialization of the ranking problem within the Information Retrieval (IR) field of study. In LeToR, machine learning algorithms are used to produce an ordered list of objects. The relative position of objects is given according to their degree of relevance or importance, depending on the problem application. A strategy present in state-of-the-art algorithms to handle LeToR is tree-based \textit{ensemble} methods. LambdaMART exemplifies this strategy and shows good results compared to other models, including deep neural networks, placing itself as state-of-the-art in LeToR. A tree-based \textit{ensemble} method not yet studied in LeToR, the Deep Forest, aims to perform deep learning without use deep neural networks. To do so, Deep Forest applies a layer-by-layer processing and an attribute transformation within the model through a \textit{ensemble} of \textit{Random Forest}. Due to the good performance shown by Deep Forest in several tasks and observing the good applicability of tree-based \textit{ensemble} methods in Learning to Rank, it is coherent to study the application of Deep Forest under the LeToR perspective. Having this general objective, the present work experimentally investigates Deep Forest aspects such as hyperparametrization, possible improvements of the original model, the model's behavior by bias and variance and the comparison with deep neural networks, all in the context of LeToR. It is expected that this investigation offers, in addition to the results of the application of Deep Forest in LeToR, an analytical view of the behavior of \textit{ensemble} models in LeToR and a comparative analysis of the results with deep neural networks. |
id |
UFG-2_9da5e108c7807eea6b2bfe188d762d6f |
---|---|
oai_identifier_str |
oai:repositorio.bc.ufg.br:tede/12087 |
network_acronym_str |
UFG-2 |
network_name_str |
Repositório Institucional da UFG |
repository_id_str |
|
spelling |
Rosa, Thierson Coutohttp://lattes.cnpq.br/4414718560764818Sousa, Daniel Xavier dehttp://lattes.cnpq.br/4603724338719739Rosa, Thierson CoutoSousa, Daniel XavierRocha, Leonardo Chaves Dutra daCanuto, Sérgio Daniel Carvalhohttp://lattes.cnpq.br/5203884944632363Rocha, Lucas Elias Cardoso2022-05-25T14:47:03Z2022-05-25T14:47:03Z2022-04-20ROCHA, L. E. C. Análise da técnica deep forest para o problema de aprendizado de ranqueamento. 2022. 74 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2022.http://repositorio.bc.ufg.br/tede/handle/tede/12087Learning to Rank (LeToR) is a specialization of the ranking problem within the Information Retrieval (IR) field of study. In LeToR, machine learning algorithms are used to produce an ordered list of objects. The relative position of objects is given according to their degree of relevance or importance, depending on the problem application. A strategy present in state-of-the-art algorithms to handle LeToR is tree-based \textit{ensemble} methods. LambdaMART exemplifies this strategy and shows good results compared to other models, including deep neural networks, placing itself as state-of-the-art in LeToR. A tree-based \textit{ensemble} method not yet studied in LeToR, the Deep Forest, aims to perform deep learning without use deep neural networks. To do so, Deep Forest applies a layer-by-layer processing and an attribute transformation within the model through a \textit{ensemble} of \textit{Random Forest}. Due to the good performance shown by Deep Forest in several tasks and observing the good applicability of tree-based \textit{ensemble} methods in Learning to Rank, it is coherent to study the application of Deep Forest under the LeToR perspective. Having this general objective, the present work experimentally investigates Deep Forest aspects such as hyperparametrization, possible improvements of the original model, the model's behavior by bias and variance and the comparison with deep neural networks, all in the context of LeToR. It is expected that this investigation offers, in addition to the results of the application of Deep Forest in LeToR, an analytical view of the behavior of \textit{ensemble} models in LeToR and a comparative analysis of the results with deep neural networks.Learning to Rank (LeToR) é uma especialização do problema de ranqueamento dentro do campo de estudo de Recuperação de Informação (RI). Em LeToR, algoritmos de aprendizado de máquina são utilizados para produzir uma lista ordenada de objetos. A posição relativa dos objetos nessa lista se dá de acordo com seus graus de relevância ou importância, conforme a aplicação do problema. Uma estratégia presente nos algoritmos do estado-da-arte para tratar LeToR são os métodos \textit{ensemble} baseados em árvores de decisão. LambdaMART exemplifica essa estratégia e mostram bons resultados comparados com outros modelos, incluindo redes neurais profundas, se colocando como estado-da-arte de LeToR. Um modelo \textit{ensemble} da literatura ainda não estudado em LeToR, o Deep Forest, busca realizar aprendizado profundo sem a utilização de redes neurais profundas. Para tal, o Deep Forest aplica um processamento camada a camada e uma transformação de atributos dentro do modelo através do \textit{ensemble} de \textit{Random Forest}. Pelo bom desempenho mostrado pelo Deep Forest em diversas tarefas e observando a boa aplicabilidade de métodos \textit{ensemble} baseados em árvores em Learning to Rank, se mostra coerente estudar a aplicação do Deep Forest sob a perspectiva de LeToR. Tendo esse objetivo geral, o presente trabalho investiga experimentalmente aspectos do Deep Forest como hiperparametrização, possíveis avanços do modelo original, o comportamento do modelo em viés e variância e a comparação com redes neurais profundas, tudo no contexto de LeToR. Essa investigação oferece, além dos resultados da aplicação do Deep Forest em LeToR, uma visão analítica do comportamento de modelos \textit{ensemble} em LeToR e uma análise comparativa dos resultados com as redes neurais profundas.Submitted by Marlene Santos (marlene.bc.ufg@gmail.com) on 2022-05-23T16:12:37Z No. of bitstreams: 2 Dissertação - Lucas Elias Cardoso Rocha - 2022.pdf: 966872 bytes, checksum: 78395746e40dae257e56b1b6b5cfb5a6 (MD5) license_rdf: 805 bytes, checksum: 4460e5956bc1d1639be9ae6146a50347 (MD5)Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2022-05-25T14:47:03Z (GMT) No. of bitstreams: 2 Dissertação - Lucas Elias Cardoso Rocha - 2022.pdf: 966872 bytes, checksum: 78395746e40dae257e56b1b6b5cfb5a6 (MD5) license_rdf: 805 bytes, checksum: 4460e5956bc1d1639be9ae6146a50347 (MD5)Made available in DSpace on 2022-05-25T14:47:03Z (GMT). No. of bitstreams: 2 Dissertação - Lucas Elias Cardoso Rocha - 2022.pdf: 966872 bytes, checksum: 78395746e40dae257e56b1b6b5cfb5a6 (MD5) license_rdf: 805 bytes, checksum: 4460e5956bc1d1639be9ae6146a50347 (MD5) Previous issue date: 2022-04-20OutroporUniversidade Federal de GoiásPrograma de Pós-graduação em Ciência da Computação (INF)UFGBrasilInstituto de Informática - INF (RG)Attribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessRecuperação de InformaçãoLearning to rankMétodos ensembleDeep forestInformation retrievalLearning to rankEnsembleCIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAOAnálise da técnica deep forest para o problema de aprendizado de ranqueamentoAnalysis of deep forest technique for learning to rank taskinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesis20500500500500261845reponame:Repositório Institucional da UFGinstname:Universidade Federal de Goiás (UFG)instacron:UFGLICENSElicense.txtlicense.txttext/plain; charset=utf-81748http://repositorio.bc.ufg.br/tede/bitstreams/ba342997-80c7-4f5f-9c6f-08eae4c3f956/download8a4605be74aa9ea9d79846c1fba20a33MD51CC-LICENSElicense_rdflicense_rdfapplication/rdf+xml; charset=utf-8805http://repositorio.bc.ufg.br/tede/bitstreams/6e3eae0c-71f9-409f-9e31-e0cc027a3340/download4460e5956bc1d1639be9ae6146a50347MD52ORIGINALDissertação - Lucas Elias Cardoso Rocha - 2022.pdfDissertação - Lucas Elias Cardoso Rocha - 2022.pdfapplication/pdf966872http://repositorio.bc.ufg.br/tede/bitstreams/f776a66f-1696-4979-a896-deed323dda32/download78395746e40dae257e56b1b6b5cfb5a6MD53tede/120872022-05-25 11:47:04.23http://creativecommons.org/licenses/by-nc-nd/4.0/Attribution-NonCommercial-NoDerivatives 4.0 Internationalopen.accessoai:repositorio.bc.ufg.br:tede/12087http://repositorio.bc.ufg.br/tedeRepositório InstitucionalPUBhttp://repositorio.bc.ufg.br/oai/requesttasesdissertacoes.bc@ufg.bropendoar:2022-05-25T14:47:04Repositório Institucional da UFG - Universidade Federal de Goiás (UFG)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 |
dc.title.pt_BR.fl_str_mv |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
dc.title.alternative.eng.fl_str_mv |
Analysis of deep forest technique for learning to rank task |
title |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
spellingShingle |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento Rocha, Lucas Elias Cardoso Recuperação de Informação Learning to rank Métodos ensemble Deep forest Information retrieval Learning to rank Ensemble CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
title_short |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
title_full |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
title_fullStr |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
title_full_unstemmed |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
title_sort |
Análise da técnica deep forest para o problema de aprendizado de ranqueamento |
author |
Rocha, Lucas Elias Cardoso |
author_facet |
Rocha, Lucas Elias Cardoso |
author_role |
author |
dc.contributor.advisor1.fl_str_mv |
Rosa, Thierson Couto |
dc.contributor.advisor1Lattes.fl_str_mv |
http://lattes.cnpq.br/4414718560764818 |
dc.contributor.advisor-co1.fl_str_mv |
Sousa, Daniel Xavier de |
dc.contributor.advisor-co1Lattes.fl_str_mv |
http://lattes.cnpq.br/4603724338719739 |
dc.contributor.referee1.fl_str_mv |
Rosa, Thierson Couto |
dc.contributor.referee2.fl_str_mv |
Sousa, Daniel Xavier |
dc.contributor.referee3.fl_str_mv |
Rocha, Leonardo Chaves Dutra da |
dc.contributor.referee4.fl_str_mv |
Canuto, Sérgio Daniel Carvalho |
dc.contributor.authorLattes.fl_str_mv |
http://lattes.cnpq.br/5203884944632363 |
dc.contributor.author.fl_str_mv |
Rocha, Lucas Elias Cardoso |
contributor_str_mv |
Rosa, Thierson Couto Sousa, Daniel Xavier de Rosa, Thierson Couto Sousa, Daniel Xavier Rocha, Leonardo Chaves Dutra da Canuto, Sérgio Daniel Carvalho |
dc.subject.por.fl_str_mv |
Recuperação de Informação Learning to rank Métodos ensemble |
topic |
Recuperação de Informação Learning to rank Métodos ensemble Deep forest Information retrieval Learning to rank Ensemble CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
dc.subject.eng.fl_str_mv |
Deep forest Information retrieval Learning to rank Ensemble |
dc.subject.cnpq.fl_str_mv |
CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
description |
Learning to Rank (LeToR) is a specialization of the ranking problem within the Information Retrieval (IR) field of study. In LeToR, machine learning algorithms are used to produce an ordered list of objects. The relative position of objects is given according to their degree of relevance or importance, depending on the problem application. A strategy present in state-of-the-art algorithms to handle LeToR is tree-based \textit{ensemble} methods. LambdaMART exemplifies this strategy and shows good results compared to other models, including deep neural networks, placing itself as state-of-the-art in LeToR. A tree-based \textit{ensemble} method not yet studied in LeToR, the Deep Forest, aims to perform deep learning without use deep neural networks. To do so, Deep Forest applies a layer-by-layer processing and an attribute transformation within the model through a \textit{ensemble} of \textit{Random Forest}. Due to the good performance shown by Deep Forest in several tasks and observing the good applicability of tree-based \textit{ensemble} methods in Learning to Rank, it is coherent to study the application of Deep Forest under the LeToR perspective. Having this general objective, the present work experimentally investigates Deep Forest aspects such as hyperparametrization, possible improvements of the original model, the model's behavior by bias and variance and the comparison with deep neural networks, all in the context of LeToR. It is expected that this investigation offers, in addition to the results of the application of Deep Forest in LeToR, an analytical view of the behavior of \textit{ensemble} models in LeToR and a comparative analysis of the results with deep neural networks. |
publishDate |
2022 |
dc.date.accessioned.fl_str_mv |
2022-05-25T14:47:03Z |
dc.date.available.fl_str_mv |
2022-05-25T14:47:03Z |
dc.date.issued.fl_str_mv |
2022-04-20 |
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.citation.fl_str_mv |
ROCHA, L. E. C. Análise da técnica deep forest para o problema de aprendizado de ranqueamento. 2022. 74 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2022. |
dc.identifier.uri.fl_str_mv |
http://repositorio.bc.ufg.br/tede/handle/tede/12087 |
identifier_str_mv |
ROCHA, L. E. C. Análise da técnica deep forest para o problema de aprendizado de ranqueamento. 2022. 74 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2022. |
url |
http://repositorio.bc.ufg.br/tede/handle/tede/12087 |
dc.language.iso.fl_str_mv |
por |
language |
por |
dc.relation.program.fl_str_mv |
20 |
dc.relation.confidence.fl_str_mv |
500 500 500 500 |
dc.relation.department.fl_str_mv |
26 |
dc.relation.cnpq.fl_str_mv |
184 |
dc.relation.sponsorship.fl_str_mv |
5 |
dc.rights.driver.fl_str_mv |
Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/ info:eu-repo/semantics/openAccess |
rights_invalid_str_mv |
Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/ |
eu_rights_str_mv |
openAccess |
dc.publisher.none.fl_str_mv |
Universidade Federal de Goiás |
dc.publisher.program.fl_str_mv |
Programa de Pós-graduação em Ciência da Computação (INF) |
dc.publisher.initials.fl_str_mv |
UFG |
dc.publisher.country.fl_str_mv |
Brasil |
dc.publisher.department.fl_str_mv |
Instituto de Informática - INF (RG) |
publisher.none.fl_str_mv |
Universidade Federal de Goiás |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFG instname:Universidade Federal de Goiás (UFG) instacron:UFG |
instname_str |
Universidade Federal de Goiás (UFG) |
instacron_str |
UFG |
institution |
UFG |
reponame_str |
Repositório Institucional da UFG |
collection |
Repositório Institucional da UFG |
bitstream.url.fl_str_mv |
http://repositorio.bc.ufg.br/tede/bitstreams/ba342997-80c7-4f5f-9c6f-08eae4c3f956/download http://repositorio.bc.ufg.br/tede/bitstreams/6e3eae0c-71f9-409f-9e31-e0cc027a3340/download http://repositorio.bc.ufg.br/tede/bitstreams/f776a66f-1696-4979-a896-deed323dda32/download |
bitstream.checksum.fl_str_mv |
8a4605be74aa9ea9d79846c1fba20a33 4460e5956bc1d1639be9ae6146a50347 78395746e40dae257e56b1b6b5cfb5a6 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFG - Universidade Federal de Goiás (UFG) |
repository.mail.fl_str_mv |
tasesdissertacoes.bc@ufg.br |
_version_ |
1798044409886932992 |