Eur. Phys. J. B 9, 559-565 (1999)
https://doi.org/10.1007/s100510050796

Recognition games

Instituto de Física, Universidade Federal Fluminense, Av. Litorânea s/n, 24210-340 Niterói, RJ, Brazil

Received: 28 October 1998
Revised: 2 December 1998
Published online: 15 June 1999

Abstract

In this work we propose an evolution model based on the competition between individuals belonging to populations of neural networks, obeying the Hopfield dynamics. The selection rule adopted relies on generalization and natural classification abilities. The results obtained through computer simulation show that these populations self-organize and evolve towards equilibrium states in the region of transition between ordered and disordered phases.