3543 articles – 5273 references  [version française]

hal-00439332, version 1

Application-driven parameter tuning methodology for dynamic neural field equations

Lucian Alecu () 1, 2, Hervé Frezza-Buet () 1

16th International Conference on Neural Information Processing - ICONIP 2009 5863/2009 (2009) 135-142

Abstract: In this paper, a method is introduced in order to qualify the performance of dynamic neural fields (DNF). The method is applied to Amari's DNF equations, in order to drive the tuning of its free parameters. An original evaluation procedure is presented, and then applied to some input evolution scenarios. Such scenarios define an applicative context, for which the parameters with the lowest evaluation are optimal.

  • 1:  SUPELEC-Campus Metz
  • SUPELEC
  • 2:  CORTEX (INRIA Lorraine - LORIA)
  • INRIA – CNRS : UMR7503 – Université Henri Poincaré - Nancy I – Université Nancy II – Institut National Polytechnique de Lorraine (INPL)
  • Domain : Computer Science/Neural and Evolutionary Computing
  • Keywords : dynamic neural field – parameters tuning
 
  • hal-00439332, version 1
  • http://hal-supelec.archives-ouvertes.fr/hal-00439332
  • oai:hal-supelec.archives-ouvertes.fr:hal-00439332
  • From: Sébastien Van Luchene <>
  • Submitted on: Monday, 7 December 2009 12:26:38
  • Updated on: Monday, 4 January 2010 11:45:07