Journal of
Engineering and Technology Research

  • Abbreviation: J. Eng. Technol. Res.
  • Language: English
  • ISSN: 2006-9790
  • DOI: 10.5897/JETR
  • Start Year: 2009
  • Published Articles: 183

Full Length Research Paper

Programmable time delays in Hopfield Neural Networks

Saad Al Shaban
Department of Computer Engineering and information Technology, University of Technology, Baghdad – Iraq.
Email: [email protected]

  • Article Number - 94CA84310525
  • Vol.2(5), pp. 86-95, May 2010
  •  Accepted: 04 December 2009
  •  Published: 31 May 2010

Abstract

 

The present work aims to use time delays generated in an artificial neural network (ANN) by developing techniques; with simple software design. This network has been modified, by adding the new Alshaban model and the hybrid model called Hopfield - Alshaban network. The feedback has been proposed as such in a recurrent network (Hopfield model). Time delay can be evaluated by proper pattern appearing, when Hopfield network has been trained in such a way as to produce a programmable time delay by changing net weights vectors, which is connected to inputs. Time delays have many applications in signal processing such as digital filtering, correlation, and in ultrasonic signal classification. These created time delays are found more preferable and reliable than that generated by conventional analog and digital systems.

 

Key words: Hopfield network, time delays, recurrent network, Hopfield-Alshaban network.

  • Articles on Google by: