A Darwinian approach to artificial neural systems
W.B. Dress, Jeff R. Knisley · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 1987
Self-organization in artificial neural systems and simulated genetic methods for determining functional parameters of such systems have been combined. This leads to a model of the genome as a self-organizing system on a supra-organism level, and to suggestions for a number of experiments to be performed on the abstract genetic space. The problem of the open system is also addressed in this model. A synthetic intelligent system is viewed as an extensible set of parameters-the evolved genome. This set also serves as a specification file for individual synthetic organisms, and algorithms for network-construction use this set as data for creating specific networks. It is this set of parameters that undergoes mutation and selection leading to a synthetic organism fulfilling the goals of the systems designer. 22 refs., 4 figs.