Neural networks with multidimensional transfer functions

Tsitouras, C

dc.contributor.author	Tsitouras, C	en
dc.date.accessioned	2014-03-01T01:18:05Z
dc.date.available	2014-03-01T01:18:05Z
dc.date.issued	2002	en
dc.identifier.issn	1045-9227	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/14796
dc.subject	Initial value problem (IVP)	en
dc.subject	Orbits	en
dc.subject	Oscillators	en
dc.subject	Runge-Kutta (RK)	en
dc.subject	Vector transfer function	en
dc.subject.classification	Computer Science, Artificial Intelligence	en
dc.subject.classification	Computer Science, Hardware & Architecture	en
dc.subject.classification	Computer Science, Theory & Methods	en
dc.subject.classification	Engineering, Electrical & Electronic	en
dc.subject.other	Algorithms	en
dc.subject.other	Backpropagation	en
dc.subject.other	Gradient methods	en
dc.subject.other	Initial value problems	en
dc.subject.other	Matrix algebra	en
dc.subject.other	Runge Kutta methods	en
dc.subject.other	Transfer functions	en
dc.subject.other	Hidden layers	en
dc.subject.other	Multidimensional transfer functions	en
dc.subject.other	Neural networks	en
dc.title	Neural networks with multidimensional transfer functions	en
heal.type	journalArticle	en
heal.identifier.primary	10.1109/72.977309	en
heal.identifier.secondary	http://dx.doi.org/10.1109/72.977309	en
heal.language	English	en
heal.publicationDate	2002	en
heal.abstract	We present a new type of neural network (NN) where the data for the input layer are the value x is an element of R , the vector y is an element of R-m associated to an initial value problem (IVP) with y'(x) = f (y(x)) and a steplength h. Then the stages of a Runge-Kutta (RK) method with trainable coefficients are used as hidden layers for the integration of the IVP using f as transfer function. We take as output two estimations y,(y) over cap of IVP at the point x+h. Training the RK method at some test problems and counting the cost of the method under the coefficients used, we may achieve coefficients that help the method to perform better at a wider class of problems.	en
heal.publisher	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC	en
heal.journalName	IEEE Transactions on Neural Networks	en
dc.identifier.doi	10.1109/72.977309	en
dc.identifier.isi	ISI:000173440100019	en
dc.identifier.volume	13	en
dc.identifier.issue	1	en
dc.identifier.spage	222	en
dc.identifier.epage	228	en