A Stochastic Evolutionary Neuron Migration Process with Emerged Hebbian Dynamics

Janne Haverinen and Juha Röning
Infotech Oulu and Department of Electrical Engineering
P.O. Box 4500, FIN-90014 University Of Oulu
FINLAND
e-mail: janne.haverinen@oulu.fi

Abstract:

In this paper, we propose a phenomenological developmental model based on a stochastic evolutionary neuron migration process (SENMP) . Employing a spatial encoding scheme with lateral interaction of neurons for artificial neural networks representing candidate solutions within a neural network  ensemble¹, neurons of the ensemble form problem-specific geometrical structures as they migrate under selective pressure. The SENMP is applied to evolve purposeful behaviors for autonomous robots and to gain new insights into the development, adaptation and plasticity in artificial neural networks. We demonstrate the feasibility and advantages of the approach by evolving a robust navigation behavior for a mobile robot. We also present some preliminary results regarding the behavior of the adapting neural network ensemble and, particularly, a phenomenon exhibiting Hebbian dynamics. 

Footnotes

... ensemble¹

In this paper, the term `neural network ensemble' is used to refer to the neural network population in order to emphasize the fact that the continuous behavior of the robot is a result of the evaluation sequence of the individual neural networks.