Genetic Algorithms (GAs) have proven to be a useful means of finding optimal or near optimal solutions to hard problems that are difficult to solve by other means. However, determining which crossover and mutation operator is best to use for a specific problem can be a complex task requiring much trial and error. Furthermore, different operators may be better suited to exploring the search space at different stages of evolution. For example, crossover and mutation operators that are more likely to disrupt fit solutions may have a less disruptive effect and better search capacity during the early stages of evolution when the average fitness is low. This paper presents an automated operator selection technique that largely overcomes these deficiencies in traditional GAs by enabling the GA to dynamically discover and utilize operators that happen to perform better at finding fitter solutions during the evolution process. We provide experimental results demonstrating the effectiveness of this approach by comparing the performance of our automatic operator selection technique with a traditional GA.
[1]
Kotaro Hirasawa,et al.
Increasing Robustness Of Genetic Algorithm
,
2002,
GECCO.
[2]
John H. Holland,et al.
Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence
,
1992
.
[3]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[4]
Hideaki Suzuki,et al.
Crossover Accelerates Evolution in GAs with a Babel-like Fitness Landscape: Mathematical Analyses
,
1999,
Evolutionary Computation.
[5]
Kanta Premji Vekaria,et al.
Selective Crossover in Genetic Algorithms: An Empirical Study
,
1998,
PPSN.
[6]
Thomas E. Gerasch,et al.
A Survey Of Parallel Algorithms For One-Dimensional Integer Knapsack Problems
,
1994
.
[7]
Michèle Sebag,et al.
Controlling Crossover through Inductive Learning
,
1994,
PPSN.