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

作     者：Debiaso Rossi, Andre Luis de Leon Ferreira de Carvalho, Andre Carlos Ponce Soares, Carlos de Souza, Bruno Feres 

作者机构：Univ Sao Paulo Inst Ciencias Matemat & Computac Sao Carlos SP Brazil Univ Porto Fac Engn INESC TEC P-4100 Oporto Portugal 

出 版 物：《NEUROCOMPUTING》 (神经计算)

年 卷 期：2014年第127卷第1期

页      面：52-64页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Fundação de Amparo à Pesquisa do Estado de São Paulo, FAPESP Fundação para a Ciência e a Tecnologia, FCT, (PTDC/EIA-EIA/098355/2008) Fundação para a Ciência e a Tecnologia, FCT Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, CAPES Conselho Nacional de Desenvolvimento Científico e Tecnológico, CNPq European Regional Development Fund, FEDER, (FCOMP - 01-0124-FEDER-022701) European Regional Development Fund, FEDER 

主　　题：Algorithm selection Meta-learning Data streams 

摘      要：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. (C) 2013 Elsevier B.V. All rights reserved.