Implementation of the self-organizing feature map on parallel computers

  • V. Demian
  • J. -C. Mignot
Part of the Lecture Notes in Computer Science book series (LNCS, volume 634)


We have introduced a new mapping and a block strategy for the SOFM allowing to increase the speed-up of the parallel implementation. This method has been implemented on a iPSC/860 MIMD parallel computer. It performs well on learning dots randomly chosen in the unit square, i.e. it learns as well as the classical algorithm. Its performances outperform the parallel implementation of the classical algorithm since the achieved speed-up is increased from 2–3 to 7. The SOFM has been implemented on a SIMD computer, namely the MasPar MPI exhibiting an almost perfect matching between Kohonen maps, the grid architecture and the synchronous programming model of the MasPar system


  1. 1.
    Chwan-Hwa Wu, Russel Hodges, Chia-Jiu Wang, “Parallelizing the Self-Organizing Feature Map on multiprocessor systems”, Parallel Computing 17, pp. 821–832, 1991.Google Scholar
  2. 2.
    A. Ultsch, H.P. Siemon, “Exploratory Data Analysis: Using Kohonen networks on Transputers”, Dpt Comp. Sc., University of Dortmund, Dec. 89.Google Scholar
  3. 3.
    Teuvo Kohonen, “Self-Organization and Associative Memory”, Springer-Verlag, Berlin, 1984.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1992

Authors and Affiliations

  • V. Demian
    • 1
  • J. -C. Mignot
    • 1
  1. 1.Laboratoire de l'Informatique du ParallèlismeEcole Normale Supérieure de LyonLyon Cedex 07France

Personalised recommendations