How a Generative Encoding Fares as Problem-Regularity Decreases
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.
KeywordsEvolution regularity modularity ANN NEAT HyperNEAT
Unable to display preview. Download preview PDF.
- 2.D’Ambrosio, D.B., Stanley, K.O.: A novel generative encoding for exploiting neural network sensor and output geometry. In: Whitley, D., Goldber, D., Cantu-Paz, E., Spector, L., Parmee, I., Beyer, H.-G. (eds.) GECCO 2007, pp. 974–981. ACM Press, New York (2007)Google Scholar
- 3.Gauci, J.J., Stanley, K.O.: Generating Large-Scale Neural Networks Through Discovering Geometric Regularities. In: Whitley, D., Goldber, D., Cantu-Paz, E., Spector, L., Parmee, I., Beyer, H.-G. (eds.) GECCO 2007, pp. 997–1004. ACM Press, New York (2007)Google Scholar
- 4.Gruau, F.: Genetic Synthesis of Boolean Neural Networks with a Cell Rewriting Developmental Process. International Workshop on Combinations of Genetic Algorithms and Neural Networks 6, 55–74 (1992)Google Scholar
- 5.Gruau, F., Whitley, D., Pyeatt, L.: A Comparison Between Cellular Encoding and Direct Encoding for Genetic Neural Networks. In: Proc. 1st Ann. Conf. on Genetic Programming 1996, pp. 81–89. MIT Press, Cambridge (1996)Google Scholar
- 7.Reisinger, J., Miikkulainen, R.: Acquiring Evolvability Through Adaptive Representations. In: Whitley, D., Goldber, D., Cantu-Paz, E., Spector, L., Parmee, I., Beyer, H.-G. (eds.) GECCO 2007, pp. 1045–1052. ACM Press, New York (2007)Google Scholar
- 8.Nolfi, S., Miglino, O., Parisi, D.: Phenotypic Plasticity in Evolving Neural Networks. In: Proc. Intl. Conf. from Perception to Action. IEEE Press, Los Alamitos (1994)Google Scholar