| Abstract |
In this talk, I will give an simple introduction on how to use REVAC for optimizing optimizers.
REVAC is an Estimation of Density Algorithm that is aimed at calibrating optimizers and
estimating the relevance of the optimizer parameters.
I will show a case study, in which we use this information to compare 120 different Genetic
Algorithms on a specific problem. Using REVAC we were able to judge about the performance
and the costs that are needed to reach this level of performance. This will allow researchers to
choose between algorithms based on their costs and benefits. Furthermore we were able to
introduce new evidence on one of the big debates in the Evolutionary Computation Community.
Which is best Mutation or Crossover? |