A Neuro-fuzzy Coding for Processing Incomplete Data: Application tothe Classification of Seismic Events

  • Authors:

  • Stéphanie Muller;Patrick Garda;Jean-Denis Muller;René Crusem;Yves Cansi

  • Affiliations:

  • Commissariat à l‘Energie Atomique, Direction des Applications Militaires, B.P. 12, F–91680, Bruyères-le-Châtel;Université Pierre et Marie Curie, Laboratoire des Instruments et Systèmes, case 252, 4, place Jussieu, F–75252 Paris Cedex 05 E-mail: mullerst@bruyeres.cea.fr;Commissariat à l‘Energie Atomique, Direction des Applications Militaires, B.P. 12, F–91680, Bruyères-le-Châtel;Commissariat à l‘Energie Atomique, Direction des Applications Militaires, B.P. 12, F–91680, Bruyères-le-Châtel;Commissariat à l‘Energie Atomique, Direction des Applications Militaires, B.P. 12, F–91680, Bruyères-le-Châtel

  • Venue:
  • Neural Processing Letters
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

This letter presents a method formodelling and processing incomplete data inconnectionist systems. The approach consists inapplying a neuro-fuzzy coding to the input data of aneural network. After an introduction to the differentkinds of imperfections, we propose a neuro-fuzzycoding in order to take incomplete data into account.We show the efficiency of this coding on the problemof the classification of seismic events. The resultsshow that a neuro-fuzzy coding of the inputs of aneural network increases the performance andclassifies incomplete data with little affect on theresults.