A guidance of data stream characterization for meta-learning

Página do item simplificado
dc.contributor.authorRossi, André Luis Debiaso [UNESP]
dc.contributor.authorDe Souza, Bruno Feres
dc.contributor.authorSoares, Carlos
dc.contributor.authorDe Carvalho, André Carlos Ponce De Leon Ferreira
dc.contributor.institutionUniversidade Estadual Paulista (Unesp)
dc.contributor.institutionUniversidade Federal Do Maranhão
dc.contributor.institutionUniversidade Do Porto
dc.contributor.institutionUniversidade de São Paulo (USP)
dc.date.accessioned2018-12-11T17:14:05Z
dc.date.available2018-12-11T17:14:05Z
dc.date.issued2017-01-01
dc.description.abstractThe problem of selecting learning algorithms has been studied by the meta-learning community for more than two decades. One of the most important task for the success of a meta-learning system is gathering data about the learning process. This data is used to induce a (meta) model able to map characteristics extracted from different data sets to the performance of learning algorithms on these data sets. These systems are built under the assumption that the data are generated by a stationary distribution, i.e., a learning algorithm will perform similarly for new data from the same problem. However, many applications generate data whose characteristics can change over time. Therefore, a suitable bias at a given time may become inappropriate at another time. Although meta-learning has been used to continuously select a learning algorithm in data streams, data characterization has received less attention in this context. In this study, we provide a set of guidelines to support the proposal of characteristics able to describe non-stationary data over time. This guidance considers both the order of arrival of the examples and the type of variables involved in the base-level learning. In addition, we analyze the influence of characteristics regarding their dependence on data morphology. Experimental results using real data streams showed the effectiveness of the proposed data characterization general scheme to support algorithm selection by meta-learning systems. Moreover, the dependent meta-features provided crucial information for the success of some meta-models.en
dc.description.affiliationUniversidade Estadual Paulista (UNESP) Campus de Itapeva, Rua Geraldo Alckmin, 519
dc.description.affiliationUniversidade Federal Do Maranhão Campus de São Luís
dc.description.affiliationINESC TEC Faculdade de Engenharia da Universidade Do Porto Universidade Do Porto
dc.description.affiliationInstituto de Ciências Matemáticas e de Computação Universidade de São Paulo
dc.description.affiliationUnespUniversidade Estadual Paulista (UNESP) Campus de Itapeva, Rua Geraldo Alckmin, 519
dc.description.sponsorshipConselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
dc.description.sponsorshipEuropean Regional Development Fund
dc.description.sponsorshipFundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
dc.description.sponsorshipFundação para a Ciência e a Tecnologia
dc.description.sponsorshipIdFAPESP: 2008/11569-6
dc.description.sponsorshipIdFundação para a Ciência e a Tecnologia: NORTE-07-0124-FEDER-000057
dc.description.sponsorshipIdFundação para a Ciência e a Tecnologia: NORTE-07-0124-FEDER-000059
dc.format.extent1015-1035
dc.identifierhttp://dx.doi.org/10.3233/IDA-160083
dc.identifier.citationIntelligent Data Analysis, v. 21, n. 4, p. 1015-1035, 2017.
dc.identifier.doi10.3233/IDA-160083
dc.identifier.issn1571-4128
dc.identifier.issn1088-467X
dc.identifier.scopus2-s2.0-85027960355
dc.identifier.urihttp://hdl.handle.net/11449/175070
dc.language.isoeng
dc.relation.ispartofIntelligent Data Analysis
dc.rights.accessRightsAcesso restrito
dc.sourceScopus
dc.subjectalgorithm selection
dc.subjectdata streams
dc.subjectFeature extraction
dc.titleA guidance of data stream characterization for meta-learningen
dc.typeArtigo
unesp.author.lattes5604829226181486[1]
unesp.author.orcid0000-0001-6388-7479[1]

Arquivos

Coleções

Artigos - Engenharia Industrial Madeireira - Itapeva