Ensembles of label noise filters: a ranking approach

Página do item simplificado
dc.contributor.authorGarcia, Luis P. F.
dc.contributor.authorLorena, Ana C. [UNIFESP]
dc.contributor.authorMatwin, Stan
dc.contributor.authorde Carvalho, Andre C. P. L. F.
dc.date.accessioned2019-07-22T15:46:50Z
dc.date.available2019-07-22T15:46:50Z
dc.date.issued2016
dc.description.abstractLabel noise can be a major problem in classification tasks, since most machine learning algorithms rely on data labels in their inductive process. Thereupon, various techniques for label noise identification have been investigated in the literature. The bias of each technique defines how suitable it is for each dataset. Besides, while some techniques identify a large number of examples as noisy and have a high false positive rate, others are very restrictive and therefore not able to identify all noisy examples. This paper investigates how label noise detection can be improved by using an ensemble of noise filtering techniques. These filters, individual and ensembles, are experimentally compared. Another concern in this paper is the computational cost of ensembles, once, for a particular dataset, an individual technique can have the same predictive performance as an ensemble. In this case the individual technique should be preferred. To deal with this situation, this study also proposes the use of meta-learning to recommend, for a new dataset, the best filter. An extensive experimental evaluation of the use of individual filters, ensemble filters and meta-learning was performed using public datasets with imputed label noise. The results show that ensembles of noise filters can improve noise filtering performance and that a recommendation system based on meta-learning can successfully recommend the best filtering technique for new datasets. A case study using a real dataset from the ecological niche modeling domain is also presented and evaluated, with the results validated by an expert.en
dc.description.affiliationUniv Sao Paulo, Inst Ciencias Matemat & Comp, Trabalhador Sao Carlense Ave 400, Sao Paulo, Brazil
dc.description.affiliationUniv Fed Sao Paulo, Inst Ciencia & Tecnol, Talim St 330, Sao Paulo, Brazil
dc.description.affiliationDalhousie Univ, Inst Big Data Analyt, Univ Ave 6050, Halifax, NS, Canada
dc.description.affiliationPolish Acad Sci, Inst Comp Sci, Warsaw, Poland
dc.description.affiliationUnifespUniv Fed Sao Paulo, Inst Ciencia & Tecnol, Talim St 330, Sao Paulo, Brazil
dc.description.sourceWeb of Science
dc.format.extent1192-1216
dc.identifierhttp://dx.doi.org/10.1007/s10618-016-0475-9
dc.identifier.citationData Mining And Knowledge Discovery. Dordrecht, v. 30, n. 5, p. 1192-1216, 2016.
dc.identifier.doi10.1007/s10618-016-0475-9
dc.identifier.issn1384-5810
dc.identifier.urihttp://repositorio.unifesp.br/handle/11600/51113
dc.identifier.wosWOS:000382010500009
dc.language.isoeng
dc.publisherSpringer
dc.relation.ispartofEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECMLPKDD)
dc.rightsinfo:eu-repo/semantics/restrictedAccess
dc.subjectLabel noiseen
dc.subjectNoise filtersen
dc.subjectEnsemble filtersen
dc.subjectNoise rankingen
dc.subjectRecommendation systemen
dc.titleEnsembles of label noise filters: a ranking approachen
dc.typeinfo:eu-repo/semantics/article
Arquivos
Coleções
ICT - Artigos