This paper presents a new algorithm for tuning the weights and bias currents of a Hopfield neural network. Generally Hopfield networks are suitable for solving combinatorial optimization problems, their main advantages being their low computational complexity and acceptable memory resource requirements. The main limit in practical use is choice of suitable coefficients to link the weight and bias current values to the conditions surrounding the problem to be solved. The algorithm presented in the paper, which is mainly based on fuzzy logic, determines these coefficients automatically thus limiting the human intervention required. The authors also define fuzzy rules that reproduce the manual experience they have acquired in determining the coefficients of the Hopfield network in a number of applications. (C) 1998 Elsevier Science B.V. All rights reserved.

Improving Hopfield neural network performance by fuzzy logic-based coefficient tuning

CAVALIERI, Salvatore;RUSSO, Marco
1998-01-01

Abstract

This paper presents a new algorithm for tuning the weights and bias currents of a Hopfield neural network. Generally Hopfield networks are suitable for solving combinatorial optimization problems, their main advantages being their low computational complexity and acceptable memory resource requirements. The main limit in practical use is choice of suitable coefficients to link the weight and bias current values to the conditions surrounding the problem to be solved. The algorithm presented in the paper, which is mainly based on fuzzy logic, determines these coefficients automatically thus limiting the human intervention required. The authors also define fuzzy rules that reproduce the manual experience they have acquired in determining the coefficients of the Hopfield network in a number of applications. (C) 1998 Elsevier Science B.V. All rights reserved.
File in questo prodotto:
File Dimensione Formato  
Improving Hopfield neural network performance.pdf

solo gestori archivio

Tipologia: Versione Editoriale (PDF)
Dimensione 1.27 MB
Formato Adobe PDF
1.27 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.11769/12499
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 10
  • ???jsp.display-item.citation.isi??? 10
social impact