On the integration of biological constraints into the evolution of artificial neural systems
Bernhard Sendhoff, Ursula Körner and Edgar Körner
The combination of evolutionary computation and artificial neural networks is increasingly reduced to a specific optimisation problem. In this work, we will propose to aim more at the integration of new principles from evolutionary neurobiology than at the mere formalisation of existing methods. We will outline several findings from neurobiology and single out the incremental development of neural systems as a principle which is worth translating into the design of artificial neural systems. Furthermore, we will argue that learning could be understood as a direct consequence of such a development process. The properties of a simulation system to test the propositions made in this paper will finally be outlined.
send email to authors

Last update 29-07-2000, all inquiries to Bernhard Sendhoff