Authors:
Martin Pyka
1
;
Tilo Kircher
1
;
Sascha Hauke
2
and
Dominik Heider
3
Affiliations:
1
University of Marburg, Germany
;
2
Technische Universität Darmstadt, Germany
;
3
University of Duisburg-Essen, Germany
Keyword(s):
Neural Networks, Artificial Development, CPPN.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Artificial Life
;
Bio-inspired Hardware and Networks
;
Cognitive Systems
;
Computational Intelligence
;
Evolution Strategies
;
Evolutionary Computing
;
Genetic Algorithms
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Soft Computing
;
Symbolic Systems
Abstract:
To study the evolution of complex nervous systems through artificial development, an encoding scheme for modeling networks is needed that reflects intrinsic properties similiar to natural encodings. Like the genetic code, a description language for simulations should indirectly encode networks, be stable but adaptable through evolution and should encode functions of neural networks through architectural design as well as single neuron configurations. We propose an indirect encoding scheme based on Compositional Pattern Producing Networks (CPPNs) to fulfill these needs. The encoding scheme uses CPPNs to generate multidimensional patterns that represent the analog to protein distributions in the development of organisms. These patterns form the template for three-dimensional neural networks, in which dendrite- and axon cones are placed in space to determine the actual connections in a spiking neural network simulation.