On the overtraining phenomenon of backpropagation neural networks

Tzafestas, SG; Dalianis, PJ; Anthopoulos, G

dc.contributor.author	Tzafestas, SG	en
dc.contributor.author	Dalianis, PJ	en
dc.contributor.author	Anthopoulos, G	en
dc.date.accessioned	2014-03-01T01:12:09Z
dc.date.available	2014-03-01T01:12:09Z
dc.date.issued	1996	en
dc.identifier.issn	0378-4754	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/11978
dc.subject	backpropagation	en
dc.subject	backpropagation neural network	en
dc.subject	Character Recognition	en
dc.subject	Dynamic Properties	en
dc.subject	Energy Function	en
dc.subject	Function Approximation	en
dc.subject	Generalization Capability	en
dc.subject	Neural Network	en
dc.subject.classification	Computer Science, Interdisciplinary Applications	en
dc.subject.classification	Computer Science, Software Engineering	en
dc.subject.classification	Mathematics, Applied	en
dc.subject.other	Algorithms	en
dc.subject.other	Approximation theory	en
dc.subject.other	Backpropagation	en
dc.subject.other	Character recognition	en
dc.subject.other	Computer simulation	en
dc.subject.other	Electric network topology	en
dc.subject.other	Learning systems	en
dc.subject.other	Performance	en
dc.subject.other	Backpropagation neural networks	en
dc.subject.other	Dynamic properties	en
dc.subject.other	Generalization capabilities	en
dc.subject.other	Network size	en
dc.subject.other	Overtraining phenomenon	en
dc.subject.other	Training set size	en
dc.subject.other	Neural networks	en
dc.title	On the overtraining phenomenon of backpropagation neural networks	en
heal.type	journalArticle	en
heal.identifier.primary	10.1016/0378-4754(95)00003-8	en
heal.identifier.secondary	http://dx.doi.org/10.1016/0378-4754(95)00003-8	en
heal.language	English	en
heal.publicationDate	1996	en
heal.abstract	A very important subject for the consolidation of neural networks is the study of their capabilities. In this paper, the relationships between network size, training set size and generalization capabilities are examined. The phenomenon of overtraining in backpropagation networks is discussed and an extension to an existing algorithm is described. The extended algorithm provides a new energy function and its advantages, such as improved plasticity and performance along with its dynamic properties, are explained. The algorithm is applied to some common problems (XOR, numeric character recognition and function approximation) and simulation results are presented and discussed.	en
heal.publisher	ELSEVIER SCIENCE BV	en
heal.journalName	Mathematics and Computers in Simulation	en
dc.identifier.doi	10.1016/0378-4754(95)00003-8	en
dc.identifier.isi	ISI:A1996UR67100002	en
dc.identifier.volume	40	en
dc.identifier.issue	5-6 SPEC. ISS.	en
dc.identifier.spage	507	en
dc.identifier.epage	521	en