MetaStream: A meta-learning based method for periodic algorithm selection in time-changing data

• Tipo: Artigo em Revista Científica Internacional
• Data: 2014

Autores

Andre Luis D Debiaso Rossi (Autor) Outra; Andre Carlos Ponce D F de Leon Ferreira de Carvalho (Autor) Outra; Carlos Soares (Autor) FEUP; Bruno Feres de Souza (Autor) Outra

Revista

Título: Neurocomputing

Vol. 127

Páginas: 52-64

ISSN: 0925-2312

Editora: Elsevier

Indexação

ISI Web of Knowledge - 57 Citações

Scopus - 71 Citações

Classificação Científica

FOS: Ciências exactas e naturais > Ciências da computação e da informação

Outras Informações

ID Authenticus: P-008-HNJ

DOI: 10.1016/j.neucom.2013.05.048

Abstract (EN): Dynamic real-world applications that generate data continuously have introduced new challenges for the machine learning community, since the concepts to be learned are likely to change over time. In such scenarios, an appropriate model at a time point may rapidly become obsolete, requiring updating or replacement. As there are several learning algorithms available, choosing one whose bias suits the current data best is not a trivial task. In this paper, we present a meta-learning based method for periodic algorithm selection in time-changing environments, named MetaStream. It works by mapping the characteristics extracted from the past and incoming data to the performance of regression models in order to choose between single learning algorithms or their combination. Experimental results for two real regression problems showed that MetaStream is able to improve the general performance of the learning system compared to a baseline method and an ensemble-based approach.

Idioma: Inglês

Tipo (Avaliação Docente): Científica

Nº de páginas: 13