Variable selection in nonlinear modeling based on RBF networks and evolutionary computation

Patrinos, P; Alexandridis, A; Ninos, K; Sarimveis, H

dc.contributor.author	Patrinos, P	en
dc.contributor.author	Alexandridis, A	en
dc.contributor.author	Ninos, K	en
dc.contributor.author	Sarimveis, H	en
dc.date.accessioned	2014-03-01T01:34:50Z
dc.date.available	2014-03-01T01:34:50Z
dc.date.issued	2010	en
dc.identifier.issn	0129-0657	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/20885
dc.subject	evolutionary computation	en
dc.subject	gas furnace data	en
dc.subject	Mackey glass data	en
dc.subject	neural networks	en
dc.subject	quantitative structure activity relationship (QSAR)	en
dc.subject	radial basis functions	en
dc.subject	Variable selection	en
dc.subject.classification	Computer Science, Artificial Intelligence	en
dc.subject.other	Evolutionary computations	en
dc.subject.other	Mackey glass data	en
dc.subject.other	Quantitative structure-activity relationships	en
dc.subject.other	Radial basis functions	en
dc.subject.other	Variable selection	en
dc.subject.other	Calculations	en
dc.subject.other	Function evaluation	en
dc.subject.other	Furnaces	en
dc.subject.other	Gas furnaces	en
dc.subject.other	Genetic algorithms	en
dc.subject.other	Glass	en
dc.subject.other	Image segmentation	en
dc.subject.other	Neural networks	en
dc.subject.other	Parameter estimation	en
dc.subject.other	Tuning	en
dc.subject.other	Radial basis function networks	en
dc.title	Variable selection in nonlinear modeling based on RBF networks and evolutionary computation	en
heal.type	journalArticle	en
heal.identifier.primary	10.1142/S0129065710002474	en
heal.identifier.secondary	http://dx.doi.org/10.1142/S0129065710002474	en
heal.language	English	en
heal.publicationDate	2010	en
heal.abstract	In this paper a novel variable selection method based on Radial Basis Function (RBF) neural networks and genetic algorithms is presented. The fuzzy means algorithm is utilized as the training method for the RBF networks, due to its inherent speed, the deterministic approach of selecting the hidden node centers and the fact that it involves only a single tuning parameter. The trade-off between the accuracy and parsimony of the produced model is handled by using Final Prediction Error criterion, based on the RBF training and validation errors, as a fitness function of the proposed genetic algorithm. The tuning parameter required by the fuzzy means algorithm is treated as a free variable by the genetic algorithm. The proposed method was tested in benchmark data sets stemming from the scientific communities of time-series prediction and medicinal chemistry and produced promising results. © 2010 World Scientific Publishing Company.	en
heal.publisher	WORLD SCIENTIFIC PUBL CO PTE LTD	en
heal.journalName	International Journal of Neural Systems	en
dc.identifier.doi	10.1142/S0129065710002474	en
dc.identifier.isi	ISI:000284649300002	en
dc.identifier.volume	20	en
dc.identifier.issue	5	en
dc.identifier.spage	365	en
dc.identifier.epage	379	en