An Investigation into the Use of Different Search Strategies with Grammatical Evolution
We present an investigation into the performance of Grammatical Evolution using a number of different search strategies, Simulated Annealing, Hill Climbing, Random Search and Genetic Algorithms. Comparative results on three different problems are examined. We analyse the nature of the search spaces presented by these problems and offer an explanation for the contrasting performance of each of the search strategies. Our results show that Genetic Algorithms provide a consistent level of performance across all three problems successfully coping with sensitivity of the system to discrete changes in the selection of productions from the associated grammar.
KeywordsGenetic Algorithm Simulated Annealing Crossover Operator Hill Climbing Symbolic Regression
Unable to display preview. Download preview PDF.
- 1.Goldberg, David E. 1989. Genetic Algorthms in Search, Optimization and Machine Learning. Addison Wesley.Google Scholar
- 2.Kirkpatrick, S., Gerlatt, C.D.Jr., and Vecchi, M.P., Optimization by Simulated Annealing, Science 220, 671–680, 1983.Google Scholar
- 3.Mitchell, M.and Holland, J.H. 1993. When will a Genetic Algorthm Outperform Hill Climbing? Technical report, Santa Fe Institute.Google Scholar
- 4.Koza, J. 1992. Genetic Programming. MIT Press.Google Scholar
- 5.Keijzer M., O’Neill M., Ryan C., Cattolico M., Babovic V. Ripple Crossover In Genetic Programming. Euro GP 2001.Google Scholar
- 6.O’Neill M., Ryan C. Grammatical Evolution. IEEE Trans. Evolutionary Computation, To appear 2001.Google Scholar
- 7.Langdon W.B., Poli R. Why Ants are Hard. Technical Report CSRP-98-4, University of Birmingham, School of Computer Science, January 1998.Google Scholar