Adaptation in natural and artificial systems
Adaptation in natural and artificial systems
Epistasis in Genetic Algorithms: An Experimental Design Perspective
Proceedings of the 6th International Conference on Genetic Algorithms
Estimating the Heritability by Decomposing the Genetic Variance
PPSN III Proceedings of the International Conference on Evolutionary Computation. The Third Conference on Parallel Problem Solving from Nature: Parallel Problem Solving from Nature
Hi-index | 0.00 |
Experimental design and evolutionary computation are different types of optimization techniques, both based on the same expectation that the achievement of a target system is improved by identifying and recombining the promising sub-components. Although the close relationship between the two techniques has long been recognized in the evolutionary computation community, past studies mostly focused on either analytical or comparative aspects of the issues. The present chapter is an attempt to combine experimental design and evolutionary computation into a single search strategy using a specific type of recombination function called a deterministic crossover operator. We first provide a brief overview of the traditional methods for experimental design as well as their connections to evolutionary computation. Then, "fair" and "greedy" sampling strategies are formulated, assuming the solution space is decomposed into two uniquely determined sub-spaces with a given deterministic crossover operator. Based on this formulation, a genetic-based implementation of fair and greedy sampling is also presented, with some illustrative experimental results.