Publication:
Using a Mahalanobis-like distance to train Radial Basis Neural Networks

Simple item page
dc.affiliation.dptoUC3M. Departamento de Informáticaes
dc.affiliation.grupoinvUC3M. Grupo de Investigación: Computación Evolutiva y Redes Neuronales (EVANNAI)es
dc.contributor.authorValls, José M.
dc.contributor.authorAler, Ricardo
dc.contributor.authorFernández, Óscar
dc.date.accessioned2009-12-14T13:27:39Z
dc.date.available2009-12-14T13:27:39Z
dc.date.issued2005-06-21
dc.descriptionProceeding of: International Work-Conference on Artificial Neural Networks (IWANN 2005)
dc.description.abstractRadial Basis Neural Networks (RBNN) can approximate any regular function and have a faster training phase than other similar neural networks. However, the activation of each neuron depends on the euclidean distance between a pattern and the neuron center. Therefore, the activation function is symmetrical and all attributes are considered equally relevant. This could be solved by altering the metric used in the activation function (i.e. using non-symmetrical metrics). The Mahalanobis distance is such a metric, that takes into account the variability of the attributes and their correlations. However, this distance is computed directly from the variance-covariance matrix and does not consider the accuracy of the learning algorithm. In this paper, we propose to use a generalized euclidean metric, following the Mahalanobis structure, but evolved by a Genetic Algorithm (GA). This GA searches for the distance matrix that minimizes the error produced by a fixed RBNN. Our approach has been tested on two domains and positive results have been observed in both cases.
dc.format.mimetypeapplication/pdf
dc.identifier.bibliographicCitationComputational intelligence and bioinspired systems, Springer, Junio 2005, p. 257-263
dc.identifier.doi10.1007/11494669_32
dc.identifier.isbn978-3-540-26208-4
dc.identifier.issn0302-9743 (Print)
dc.identifier.issn1611-3349 (Online)
dc.identifier.publicationfirstpage257
dc.identifier.publicationlastpage263
dc.identifier.publicationtitleComputational intelligence and bioinspired systems
dc.identifier.urihttps://hdl.handle.net/10016/6033
dc.language.isoeng
dc.publisherSpringer
dc.relation.eventdate8 to 10 June 2005
dc.relation.eventnumber8
dc.relation.eventplaceVilanova i la Geltrú (Barcelona, Spain)
dc.relation.eventtitleInternational Work-Conference on Artificial Neural Networks (IWANN)
dc.relation.ispartofseriesLecture notes in computer science, vol. 3512
dc.relation.publisherversionhttp://dx.doi.org/10.1007/11494669_32
dc.rights© Springer
dc.rights.accessRightsopen access
dc.subject.otherRadial Basis Neural Networks
dc.titleUsing a Mahalanobis-like distance to train Radial Basis Neural Networks
dc.typeconference paper*
dspace.entity.typePublication
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
using_aler_IWANN_2005_ps.pdf
Size:
369.35 KB
Format:
Adobe Portable Document Format
Download
Collections
DI - GCERN - Capítulos de Monografías
DI - GCERN - Comunicaciones en Congresos y otros eventos