Title Evolutionary product-unit neural networks classifiers
Authors MARTÍNEZ ESTUDILLO, FRANCISCO JOSÉ, Hervás-Martínez C., Gutiérrez P.A., MARTÍNEZ ESTUDILLO, ALFONSO CARLOS, MARTÍNEZ ESTUDILLO, FRANCISCO JOSÉ, MARTÍNEZ ESTUDILLO, ALFONSO CARLOS
External publication No
Means Neurocomputing
Scope Article
Nature Científica
JCR Quartile 3
SJR Quartile 2
JCR Impact 1.23400
SJR Impact 0.51400
Area International
Web https://www.scopus.com/inward/record.uri?eid=2-s2.0-55349128771&doi=10.1016%2fj.neucom.2007.11.019&partnerID=40&md5=e254691dd2ee0323cbadba4305481b65
Publication date 01/01/2008
ISI 000261643700060
Scopus Id 2-s2.0-55349128771
DOI 10.1016/j.neucom.2007.11.019
Abstract This paper proposes a classification method based on a special class of feed-forward neural network, namely product-unit neural networks. Product-units are based on multiplicative nodes instead of additive ones, where the nonlinear basis functions express the possible strong interactions between variables. We apply an evolutionary algorithm to determine the basic structure of the product-unit model and to estimate the coefficients of the model. We use softmax transformation as the decision rule and the cross-entropy error function because of its probabilistic interpretation. The approach can be seen as nonlinear multinomial logistic regression where the parameters are estimated using evolutionary computation. The empirical and specific multiple comparison statistical test results, carried out over several benchmark data sets and a complex real microbial Listeria growth/no growth problem, show that the proposed model is promising in terms of its classification accuracy and the number of the model coefficients, yielding a state-of-the-art performance. © 2008 Elsevier B.V. All rights reserved.
Keywords Evolutionary algorithms; Feedforward neural networks; Ketones; Probability density function; Statistical tests; Trees (mathematics); Vegetation; Basic structures; Benchmark datums; Classification; Cla
Universidad Loyola members