Aprendizado supervisionado usando redes neurais construtivas
Autor(a) principal: | |
---|---|
Data de Publicação: | 2006 |
Tipo de documento: | Dissertação |
Idioma: | por |
Título da fonte: | Repositório Institucional da UFSCAR |
Texto Completo: | https://repositorio.ufscar.br/handle/ufscar/323 |
Resumo: | Constructive neural learning is a neural learning model that does not assume a fixed network topology before training begins. The main characteristic of this learning model is the dynamic construction of the network s hidden layers that occurs simultaneously with training. This work investigates three topics related to constructive neural learning namely algorithms for training an individual TLU, constructive neural algorithms for two class problems and constructive neural algorithms for multiclass problems. The first research topic is approached by discussing a few TLU training algorithms, namely Perceptron, Pocket, Thermal, Modified Thermal, MinOver and BCP. This work approaches constructive neural learning for two class classification tasks by initially reviewing Tower, Pyramid, Tiling and Upstart algorithms, aiming at their multiclass versions. Next five constructive neural algorithms namely Shift, Offset, PTI, Perceptron Cascade and Sequential are investigated and two hybrid algorithms are proposed: Hybrid Tiling, that does not restrict the TLU s training to only one algorithm and the OffTiling, a collaborative approach based on Tiling and Offset. Multiclass constructive neural learning was approached by investigating TLUs training algorithms that deal with multiclass as well as by investigating multiclass versions of Tower, Pyramid, Tiling, Upstart and Perceptron Cascade. This research work also describes an empirical evaluation of all the investigated algorithms conducted using several knowledge domains. Results are discussed and analyzed. |
id |
SCAR_30a153475709651338d9d81d06cbab10 |
---|---|
oai_identifier_str |
oai:repositorio.ufscar.br:ufscar/323 |
network_acronym_str |
SCAR |
network_name_str |
Repositório Institucional da UFSCAR |
repository_id_str |
4322 |
spelling |
Bertini Junior, João RobertoNicoletti, Maria do Carmohttp://genos.cnpq.br:12010/dwlattes/owa/prc_imp_cv_int?f_cod=K4787728A5http://buscatextual.cnpq.br/buscatextual/visualizacv.do?id=E4418029e9baa0de-b3c9-4d33-b812-cbe9e1fe0b802016-06-02T19:05:19Z2007-07-112016-06-02T19:05:19Z2006-05-25BERTINI JUNIOR, João Roberto. Aprendizado supervisionado usando redes neurais construtivas.. 2006. 228 f. Dissertação (Mestrado em Ciências Exatas e da Terra) - Universidade Federal de São Carlos, São Carlos, 2006.https://repositorio.ufscar.br/handle/ufscar/323Constructive neural learning is a neural learning model that does not assume a fixed network topology before training begins. The main characteristic of this learning model is the dynamic construction of the network s hidden layers that occurs simultaneously with training. This work investigates three topics related to constructive neural learning namely algorithms for training an individual TLU, constructive neural algorithms for two class problems and constructive neural algorithms for multiclass problems. The first research topic is approached by discussing a few TLU training algorithms, namely Perceptron, Pocket, Thermal, Modified Thermal, MinOver and BCP. This work approaches constructive neural learning for two class classification tasks by initially reviewing Tower, Pyramid, Tiling and Upstart algorithms, aiming at their multiclass versions. Next five constructive neural algorithms namely Shift, Offset, PTI, Perceptron Cascade and Sequential are investigated and two hybrid algorithms are proposed: Hybrid Tiling, that does not restrict the TLU s training to only one algorithm and the OffTiling, a collaborative approach based on Tiling and Offset. Multiclass constructive neural learning was approached by investigating TLUs training algorithms that deal with multiclass as well as by investigating multiclass versions of Tower, Pyramid, Tiling, Upstart and Perceptron Cascade. This research work also describes an empirical evaluation of all the investigated algorithms conducted using several knowledge domains. Results are discussed and analyzed.Aprendizado neural construtivo é um modelo de aprendizado neural que não pressupõe a definição de uma topologia de rede fixada antes do início do treinamento. A principal característica deste modelo de aprendizado é a construção dinâmica das camadas intermediárias da rede, à medida que vão sendo necessárias ao seu treinamento. Este trabalho investiga três frentes de pesquisas com relação ao aprendizado neural construtivo, a saber, algoritmos para o treinamento de TLUs, algoritmos neurais construtivos para problemas que envolvem duas classes e algoritmos neurais construtivos para o tratamento de problemas multiclasses. Com relação à primeira frente de pesquisa os algoritmos discutidos para o treinamento de TLUs são o Perceptron, o Pocket, o PMR, o Thermal, o Thermal Modificado, o MinOver e o BPC. Na frente de pesquisa relativa ao aprendizado neural construtivo para duas classes são revistos os algoritmos Tower, Pyramid, Tiling e Upstart, para que as versões multiclasses desses algoritmos possam ser tratadas. São investigados os algoritmos neurais construtivos Shift, Offset, PTI, Perceptron Cascade e Sequential e propostos dois algoritmos híbridos: o Tiling Híbrido, que não restringe o treinamento de TLUs a um único algoritmo e o OffTiling que agrega os algoritmos Tiling e Offset. A frente que focaliza o aprendizado neural construtivo multiclasse investiga os algoritmos para o treinamento de TLUs quando o problema envolvido apresentar mais que duas classes bem como apresenta e discute as versões multiclasses dos algoritmos Tower, Pyramid, Tiling, Upstart e Perceptron Cascade. O trabalho descreve uma avaliação empírica dos algoritmos investigados, em vários domínios de conhecimento bem como discute e analisa os resultados obtidos.Financiadora de Estudos e Projetosapplication/pdfporUniversidade Federal de São CarlosPrograma de Pós-Graduação em Ciência da Computação - PPGCCUFSCarBRRedes neurais (Computação)Inteligência artificialAprendizado do computadorReconhecimento de padrõesCIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAOAprendizado supervisionado usando redes neurais construtivasinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesis-1-1a06526c6-28b8-482a-b64f-3655cf6d3c38info:eu-repo/semantics/openAccessreponame:Repositório Institucional da UFSCARinstname:Universidade Federal de São Carlos (UFSCAR)instacron:UFSCARORIGINALDissJRBJ.pdfapplication/pdf1580108https://repositorio.ufscar.br/bitstream/ufscar/323/1/DissJRBJ.pdf0456aef8508685d7e69da2661dc5877bMD51TEXTDissJRBJ.pdf.txtDissJRBJ.pdf.txtExtracted texttext/plain512982https://repositorio.ufscar.br/bitstream/ufscar/323/2/DissJRBJ.pdf.txtab54ed1ceb2b0fdda2b9658d17a9bc39MD52THUMBNAILDissJRBJ.pdf.jpgDissJRBJ.pdf.jpgIM Thumbnailimage/jpeg6652https://repositorio.ufscar.br/bitstream/ufscar/323/3/DissJRBJ.pdf.jpgf5df7c0f7439d4c7bbf5f19328965fffMD53ufscar/3232023-09-18 18:30:37.198oai:repositorio.ufscar.br:ufscar/323Repositório InstitucionalPUBhttps://repositorio.ufscar.br/oai/requestopendoar:43222023-09-18T18:30:37Repositório Institucional da UFSCAR - Universidade Federal de São Carlos (UFSCAR)false |
dc.title.por.fl_str_mv |
Aprendizado supervisionado usando redes neurais construtivas |
title |
Aprendizado supervisionado usando redes neurais construtivas |
spellingShingle |
Aprendizado supervisionado usando redes neurais construtivas Bertini Junior, João Roberto Redes neurais (Computação) Inteligência artificial Aprendizado do computador Reconhecimento de padrões CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
title_short |
Aprendizado supervisionado usando redes neurais construtivas |
title_full |
Aprendizado supervisionado usando redes neurais construtivas |
title_fullStr |
Aprendizado supervisionado usando redes neurais construtivas |
title_full_unstemmed |
Aprendizado supervisionado usando redes neurais construtivas |
title_sort |
Aprendizado supervisionado usando redes neurais construtivas |
author |
Bertini Junior, João Roberto |
author_facet |
Bertini Junior, João Roberto |
author_role |
author |
dc.contributor.authorlattes.por.fl_str_mv |
http://buscatextual.cnpq.br/buscatextual/visualizacv.do?id=E4418029 |
dc.contributor.author.fl_str_mv |
Bertini Junior, João Roberto |
dc.contributor.advisor1.fl_str_mv |
Nicoletti, Maria do Carmo |
dc.contributor.advisor1Lattes.fl_str_mv |
http://genos.cnpq.br:12010/dwlattes/owa/prc_imp_cv_int?f_cod=K4787728A5 |
dc.contributor.authorID.fl_str_mv |
e9baa0de-b3c9-4d33-b812-cbe9e1fe0b80 |
contributor_str_mv |
Nicoletti, Maria do Carmo |
dc.subject.por.fl_str_mv |
Redes neurais (Computação) Inteligência artificial Aprendizado do computador Reconhecimento de padrões |
topic |
Redes neurais (Computação) Inteligência artificial Aprendizado do computador Reconhecimento de padrões CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
dc.subject.cnpq.fl_str_mv |
CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO |
description |
Constructive neural learning is a neural learning model that does not assume a fixed network topology before training begins. The main characteristic of this learning model is the dynamic construction of the network s hidden layers that occurs simultaneously with training. This work investigates three topics related to constructive neural learning namely algorithms for training an individual TLU, constructive neural algorithms for two class problems and constructive neural algorithms for multiclass problems. The first research topic is approached by discussing a few TLU training algorithms, namely Perceptron, Pocket, Thermal, Modified Thermal, MinOver and BCP. This work approaches constructive neural learning for two class classification tasks by initially reviewing Tower, Pyramid, Tiling and Upstart algorithms, aiming at their multiclass versions. Next five constructive neural algorithms namely Shift, Offset, PTI, Perceptron Cascade and Sequential are investigated and two hybrid algorithms are proposed: Hybrid Tiling, that does not restrict the TLU s training to only one algorithm and the OffTiling, a collaborative approach based on Tiling and Offset. Multiclass constructive neural learning was approached by investigating TLUs training algorithms that deal with multiclass as well as by investigating multiclass versions of Tower, Pyramid, Tiling, Upstart and Perceptron Cascade. This research work also describes an empirical evaluation of all the investigated algorithms conducted using several knowledge domains. Results are discussed and analyzed. |
publishDate |
2006 |
dc.date.issued.fl_str_mv |
2006-05-25 |
dc.date.available.fl_str_mv |
2007-07-11 2016-06-02T19:05:19Z |
dc.date.accessioned.fl_str_mv |
2016-06-02T19:05:19Z |
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 |
BERTINI JUNIOR, João Roberto. Aprendizado supervisionado usando redes neurais construtivas.. 2006. 228 f. Dissertação (Mestrado em Ciências Exatas e da Terra) - Universidade Federal de São Carlos, São Carlos, 2006. |
dc.identifier.uri.fl_str_mv |
https://repositorio.ufscar.br/handle/ufscar/323 |
identifier_str_mv |
BERTINI JUNIOR, João Roberto. Aprendizado supervisionado usando redes neurais construtivas.. 2006. 228 f. Dissertação (Mestrado em Ciências Exatas e da Terra) - Universidade Federal de São Carlos, São Carlos, 2006. |
url |
https://repositorio.ufscar.br/handle/ufscar/323 |
dc.language.iso.fl_str_mv |
por |
language |
por |
dc.relation.confidence.fl_str_mv |
-1 -1 |
dc.relation.authority.fl_str_mv |
a06526c6-28b8-482a-b64f-3655cf6d3c38 |
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.publisher.none.fl_str_mv |
Universidade Federal de São Carlos |
dc.publisher.program.fl_str_mv |
Programa de Pós-Graduação em Ciência da Computação - PPGCC |
dc.publisher.initials.fl_str_mv |
UFSCar |
dc.publisher.country.fl_str_mv |
BR |
publisher.none.fl_str_mv |
Universidade Federal de São Carlos |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFSCAR instname:Universidade Federal de São Carlos (UFSCAR) instacron:UFSCAR |
instname_str |
Universidade Federal de São Carlos (UFSCAR) |
instacron_str |
UFSCAR |
institution |
UFSCAR |
reponame_str |
Repositório Institucional da UFSCAR |
collection |
Repositório Institucional da UFSCAR |
bitstream.url.fl_str_mv |
https://repositorio.ufscar.br/bitstream/ufscar/323/1/DissJRBJ.pdf https://repositorio.ufscar.br/bitstream/ufscar/323/2/DissJRBJ.pdf.txt https://repositorio.ufscar.br/bitstream/ufscar/323/3/DissJRBJ.pdf.jpg |
bitstream.checksum.fl_str_mv |
0456aef8508685d7e69da2661dc5877b ab54ed1ceb2b0fdda2b9658d17a9bc39 f5df7c0f7439d4c7bbf5f19328965fff |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFSCAR - Universidade Federal de São Carlos (UFSCAR) |
repository.mail.fl_str_mv |
|
_version_ |
1802136243591446528 |