Computing convex-layers by a multi-layer self-organizing neural network

Abstract

A multi-layer self-organizing neural network model has been proposed for computation of the convex-layers of a given set of planar points. Computation of convex-layers has been found to be useful in pattern recognition and in statistics. The proposed network architecture evolves in such a manner that it adapts itself to the hull-vertices of the convex-layers in the required order. Time complexity of the proposed model is also discussed. © Springer-Verlag Berlin Heidelberg 2004.