ABSTRACT
Genetic algorithms have particular potential as a tool for optimization when the evaluation function is noisy. Several types of genetic algorithms are compared against a mutation-driven stochastic hill-climbing algorithm on a standard set of benchmark functions which have had Gaussian noise added to them. Different criteria for judging the effectiveness of the search are also considered. The genetic algorithms used in these comparisons include an elitist simple genetic algorithm, the CHC adaptive search algorithm, and the delta coding genetic algorithm. The chapter describes several hybrid genetic algorithms and compares on a very large and noisy seismic data imaging problem. J. M Fitzpatrick and J. Grefenstette have provided empirical and analytic evidence suggesting that genetic algorithms exhibit a certain tolerance for noise. Additional performance comparisons involving a version of the delta coding algorithm combined with a local search algorithm were carried out for a seismic data interpretation problem known to have a noisy objective function.
