How a generative encoding fares as problem-regularity decreases

Author(s): 
Clune J
Ofria C
Pennock RT
Year: 
2008
Abstract: 

It has been shown that generative representations, which allow the reuse of code, perform well on problems with high regularity (i.e. where a phenotypic motif must be repeated many times). To date, however, generative representations have not been tested on irregular problems. It is unknown how they will fare on problems with intermediate and low amounts of regularity. This paper compares a generative representation to a direct representation on problems that range from having multiple types of regularity to one that is completely irregular. As the regularity of the problem decreases, the performance of the generative representation degrades to, and then underperforms, the direct encoding. The degradation is not linear, however, yet tends to be consistent for different types of problem regularity. Furthermore, if the regularity of each type is sufficiently high, the generative encoding can simultaneously exploit different types of regularities.

Pub. Info: 
Proceedings of the 10th International Conference on Parallel Problem Solving From Nature. 358-367
BibTeX: 

@incollection{
year={2008},
isbn={978-3-540-87699-1},
booktitle={Parallel Problem Solving from Nature – PPSN X},
volume={5199},
series={Lecture Notes in Computer Science},
editor={Rudolph, Günter and Jansen, Thomas and Lucas, Simon and Poloni, Carlo and Beume, Nicola},
doi={10.1007/978-3-540-87700-4_36},
title={How a Generative Encoding Fares as Problem-Regularity Decreases},
url={http://dx.doi.org/10.1007/978-3-540-87700-4_36},
publisher={Springer Berlin Heidelberg},
keywords={Evolution; regularity; modularity; ANN; NEAT; HyperNEAT},
author={Clune, Jeff and Ofria, Charles and Pennock, RobertT.},
pages={358-367},
language={English}
}