Title: Minimal distance methods

  • Objective: combine the minimal distance methods with neural methods
  • Participants: Włodzisław Duch, Karol Grudzinski
  • Dates: 1.11.1996 -present
  • Result(s):
    1. Duch W, Neural minimal distance methods, Third Conference on Neural Networks and Their Applications, Kule, October 1997
    2. Duch W, Grudzinski K and Diercksen Geerd H.F, Minimal distance neural methods. WCCI'98 (in print)
  • Problems:

    • General theory is quite rich; work out details of parametrization for the minimal distance methods, connect it with the local coordinate system approach. Test neural k-NN and similar methods.

    Use Finsler geometry ?

    Neural networks from the point of view of minimal distance methods and local coordinate systems - test new models of non-cartesian neural networks.

    Neural decision trees - hierarchical clusterization.

    Working log: (local accsess only)