Developmental neural heterogeneity through coarse-coding regulation

Authors:
Jekanthan Thangavelautham;Gabriele M. T. D'Eleuterio
Affiliations:
Institute for Aerospace Studies, University of Toronto, Toronto, Ontario, Canada;Institute for Aerospace Studies, University of Toronto, Toronto, Ontario, Canada
Venue:
ECAL'07 Proceedings of the 9th European conference on Advances in artificial life
Year:
2007

Citing 8
Cited 0

Automatic definition of modular neural networks

Adaptive Behavior
A developmental model for the evolution of artificial neural networks

Artificial Life
Evolutionary Robotics: The Biology,Intelligence,and Technology

Evolutionary Robotics: The Biology,Intelligence,and Technology
Evolving Robot Behaviours with Diffusing Gas Networks

Proceedings of the First European Workshop on Evolutionary Robotics
Co-evolving recurrent neurons learn deep memory POMDPs

GECCO '05 Proceedings of the 7th annual conference on Genetic and evolutionary computation
Long Short-Term Memory

Neural Computation
A morphogenetic evolutionary system: phylogenesis of the POEtic circuit

ICES'03 Proceedings of the 5th international conference on Evolvable systems: from biology to hardware
A coarse-coding framework for a gene-regulatory-based artificial neural tissue

ECAL'05 Proceedings of the 8th European conference on Advances in Artificial Life

Quantified Score

Hi-index	0.00

Visualization

Abstract

A coarse-coding regulatory model that facilitates neural heterogeneity through a morphogenetic process is presented. The model demonstrates cellular and tissue extensibility through ontogeny, resulting in the emergence of neural heterogeneity, use of gated memory and multistate functionality in a Artificial Neural Tissue framework. In each neuron, multiple networks of proteins compete and cooperate for representation through a coarse-coding regulatory scheme. Intracellular competition and cooperation is found to better facilitate evolutionary adaptability and result in simpler solutions than does the use of homogeneous binary neurons. The emergent use of gated memory functions within this cell model is found to be more effective than recurrent architectures for memory-dependent variants of the unlabeled sign-following robotic task.