Using Statistical Techniques to Predict GA Performance

Authors:
Rafael Nogueras;Carlos Cotta
Affiliations:
-;-
Venue:
IWANN '01 Proceedings of the 6th International Work-Conference on Artificial and Natural Neural Networks: Connectionist Models of Neurons, Learning Processes and Artificial Intelligence-Part I
Year:
2001

Citing 4
Cited 0

Optimal Mutation Rates in Genetic Search

Proceedings of the 5th International Conference on Genetic Algorithms
Empirical investigation of multiparent recombination operators in evolution strategies

Evolutionary Computation
Genetic forma recombination in permutation flowshop problems

Evolutionary Computation
No free lunch theorems for optimization

IEEE Transactions on Evolutionary Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

The design of models efficiently predicting the performance of a particular genetic algorithm on a given fitness landscape is a very important issue of practical interest. Virtual Genetic Algorithms (VGAs) constitute a statistical approach aimed at this objective. This work describes different improvements to the standard VGA model. These improvements are based on the use of a more representative dataset for the statistical analysis, the partitioning of this dataset into separate prediction models, and the utilization of a more sophisticated statistical model to grasp the destribution of fitnesses. The empirical evaluation of this enhanced model shows a more accurate fitness prediction. Furthermore, fast qualitative assessment of parameter changes is shown to be possible.