 |
    |
Voronoi Diagrams Based Function Identification
Carlos Kavka1 and Marc Schoenauer2
1LIDIC, Departamento de Informática,
Universidad Nacional de San Luis
D5700HHW, San Luis, Argentina
ckavka@unsl.edu.ar
2Projet Fractales,
INRIA Rocquencourt
BP 105,
78153 Le Chesnay Cedex, France
Marc.Schoenauer@inria.fr
Abstract. Evolutionary algorithms have been applied to function identification problems with great success. This paper presents an approach in which the individuals represent a partition of the input space in Voronoi regions together with a set of local functions associated to each one of these regions. In this way, the solution corresponds to a combination of local functions over a spatial structure topologically represented by a Voronoi diagram. Experiments show that the evolutionary algorithm can successfully evolve both the partition of the input space and the parameters of the local functions in simple problems.
LNCS 2723, p. 1089 ff.
lncs@springer.de
© Springer-Verlag Berlin Heidelberg 2003