When Evolving Populations is Better than Coevolving Individuals: The Blind Mice Problem
Proceedings of the Eighteenth International Joint Conference on Artificial Intelligence. Pages 647-651.
Abstract
This . paper is about the evolutionary design of multi-agent systems. An important part of recent research in this domain has been focusing on collaborative coevolutionary methods. We expose possible drawbacks of these methods, and show that for a non-trivial problem called the "blind mice" problem, a classical GA approach in which whole populations are evaluated, selected and crossed together (with a few tweaks) finds an elegant and non-intuitive solution more efficiently than cooperative coevolution. The difference in efficiency grows with the number of agents within the simulation. We propose an explanation for this poorer performance of cooperative coevolution, based on the intrinsic fragility of the evaluation process. This explanation is supported by theoretical and experimental arguments.
@inproceedings{ijcai2003p95,
title = {When Evolving Populations is Better than Coevolving Individuals: The Blind Mice Problem},
author = {Miconi, Thomas},
booktitle = {Proceedings of the Eighteenth International Joint Conference on
Artificial Intelligence, {IJCAI-03}},
publisher = {International Joint Conferences on Artificial Intelligence},
editor = {Georg Gottlob},
pages = {647--651},
year = {2003},
url = {https://ijcai.org/proceedings/2003/95},
}
TY - CPAPER TI - When Evolving Populations is Better than Coevolving Individuals: The Blind Mice Problem AU - Miconi, Thomas T2 - Proceedings of the Eighteenth International Joint Conference on Artificial Intelligence, IJCAI-03 PB - International Joint Conferences on Artificial Intelligence PY - 2003 SP - 647 EP - 651 UR - https://ijcai.org/proceedings/2003/95 ER -Copied
Machine-readable versions of this paper, for tools and AI agents:
