Please use this identifier to cite or link to this item: http://repositorio.unicamp.br/jspui/handle/REPOSIP/341993
Full metadata record
DC FieldValueLanguage
dc.contributor.CRUESPUNIVERSIDADE ESTADUAL DE CAMPINASpt_BR
dc.identifier.isbn978-1-5386-7769-8pt_BR
dc.contributor.authorunicampGrim, Luís Fernando Lopes-
dc.contributor.authorunicampGradvohl, André Leon Sampaio-
dc.typeArtigopt_BR
dc.titleHigh-performance ensembles of online sequential extreme learning machine for regression and time series forecastingpt_BR
dc.contributor.authorGrim, Luís Fernando L.-
dc.contributor.authorGradvohl, André Leon S.-
dc.subjectLinguagem de programação (Computadores)pt_BR
dc.subjectMineração de dadospt_BR
dc.subjectAprendizado de máquinapt_BR
dc.subjectMultitarefa (Computação)pt_BR
dc.subjectAnálise de séries temporaispt_BR
dc.subjectFluxo de dados (Computadores)pt_BR
dc.subject.otherlanguageProgramming languages (Electronic computers)pt_BR
dc.subject.otherlanguageData miningpt_BR
dc.subject.otherlanguageMachine learningpt_BR
dc.subject.otherlanguageMultitasking (Computing)pt_BR
dc.subject.otherlanguageTime-series analysispt_BR
dc.subject.otherlanguageData flow computingpt_BR
dc.description.abstractEnsembles of Online Sequential Extreme Learning Machine algorithm are suitable for forecasting Data Streams with Concept Drifts. Nevertheless, data streams forecasting require high-performance implementations due to the high incoming samples rate. In this work, we proposed to tune-up three ensembles, which operates with the Online Sequential Extreme Learning Machine, using high-performance techniques. We reim-plemented them in the C programming language with Intel MKL and MPI libraries. The Intel MKL provides functions that explore the multithread features in multicore CPUs, which expands the parallelism to multiprocessors architectures. The MPI allows us to parallelize tasks with distributed memory on several processes, which can be allocated within a single computational node, or spread over several nodes. In summary, our proposal consists of a two-level parallelization, where we allocated each ensemble model into an MPI process, and we parallelized the internal functions of each model in a set of threads through Intel MKL. Thus, the objective of this work is to verify if our proposals provide a significant improvement in execution time when compared to the respective conventional serial approaches. For the experiments, we used a synthetic and a real dataset. Experimental results showed that, in general, the high-performance ensembles improve the execution time, when compared with its serial version, performing up to 10-fold fasterpt_BR
dc.relation.ispartofSymposium on computer architecture and high performance computing. proceedingspt_BR
dc.publisher.cityWashingtonpt_BR
dc.publisher.countryEstados Unidospt_BR
dc.publisherIEEE Computer Societypt_BR
dc.date.issued2019-
dc.date.monthofcirculationFeb.pt_BR
dc.language.isoengpt_BR
dc.rightsFechadopt_BR
dc.sourceSCOPUSpt_BR
dc.identifier.issn1550-6533pt_BR
dc.identifier.doi10.1109/CAHPC.2018.8645863pt_BR
dc.identifier.urlhttps://ieeexplore.ieee.org/abstract/document/8645863pt_BR
dc.date.available2020-05-22T14:16:29Z-
dc.date.accessioned2020-05-22T14:16:29Z-
dc.description.provenanceSubmitted by Susilene Barbosa da Silva (susilene@unicamp.br) on 2020-05-22T14:16:29Z No. of bitstreams: 0. Added 1 bitstream(s) on 2020-08-27T19:17:53Z : No. of bitstreams: 1 2-s2.0-85063145523.pdf: 299503 bytes, checksum: a236f4d953ae23acbd585924937972d3 (MD5)en
dc.description.provenanceMade available in DSpace on 2020-05-22T14:16:29Z (GMT). No. of bitstreams: 0 Previous issue date: 2019en
dc.identifier.urihttp://repositorio.unicamp.br/jspui/handle/REPOSIP/341993-
dc.description.conferencenome30th international symposium on computer architecture and high performance computing (SBAC-PAD)pt_BR
dc.contributor.departmentSem informaçãopt_BR
dc.contributor.departmentSem informaçãopt_BR
dc.contributor.unidadeFaculdade de Tecnologiapt_BR
dc.contributor.unidadeFaculdade de Tecnologiapt_BR
dc.subject.keywordE-learningpt_BR
dc.subject.keywordRegressionpt_BR
dc.subject.keywordTime series forecastingpt_BR
dc.subject.keywordForecastingpt_BR
dc.identifier.source2-s2.0-85063145523pt_BR
dc.creator.orcid0000-0002-1221-4095pt_BR
dc.creator.orcid0000-0002-6520-9740pt_BR
dc.type.formArtigo de eventopt_BR
dc.identifier.articleid18473263pt_BR
Appears in Collections:FT - Artigos e Outros Documentos

Files in This Item:
File Description SizeFormat 
2-s2.0-85063145523.pdf292.48 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.