Search Space Restriction of Neuro-evolution through Constrained Modularization of Neural Networks

Christian W. Rempis, Frank Pasemann

2010

Abstract

Evolving recurrent neural networks for behavior control of robots equipped with larger sets of sensors and actuators is difficult due to the large search spaces that come with the larger number of input and output neurons. We propose constrained modularization as a novel technique to reduce the search space for such evolutions. Appropriate neural networks are divided manually into logically and spatially related neuro-modules based on domain knowledge of the targeted problem. Then constraint functions are applied to these neuro-modules to force the compliance of user defined restrictions and relations. For neuro-modules this will facilitate complex symmetries and other spatial relations, local processing of related sensors and actuators, the reuse of functional neuro-modules, fine control of synaptic connections, and a non-destructive crossover operator. With an implementation of this so called ICONE method several behaviors for non-trivial robots have already been evolved successfully.

Download


Paper Citation


in Harvard Style

W. Rempis C. and Pasemann F. (2010). Search Space Restriction of Neuro-evolution through Constrained Modularization of Neural Networks . In Proceedings of the 6th International Workshop on Artificial Neural Networks and Intelligent Information Processing - Volume 1: Workshop ANNIIP, (ICINCO 2010) ISBN 978-989-8425-03-4, pages 13-22. DOI: 10.5220/0003026200130022


in Bibtex Style

@conference{workshop anniip10,
author={Christian W. Rempis and Frank Pasemann},
title={Search Space Restriction of Neuro-evolution through Constrained Modularization of Neural Networks},
booktitle={Proceedings of the 6th International Workshop on Artificial Neural Networks and Intelligent Information Processing - Volume 1: Workshop ANNIIP, (ICINCO 2010)},
year={2010},
pages={13-22},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0003026200130022},
isbn={978-989-8425-03-4},
}


in EndNote Style

TY - CONF
JO - Proceedings of the 6th International Workshop on Artificial Neural Networks and Intelligent Information Processing - Volume 1: Workshop ANNIIP, (ICINCO 2010)
TI - Search Space Restriction of Neuro-evolution through Constrained Modularization of Neural Networks
SN - 978-989-8425-03-4
AU - W. Rempis C.
AU - Pasemann F.
PY - 2010
SP - 13
EP - 22
DO - 10.5220/0003026200130022