Abstract
Automatic algorithm configurators are important practical tools for improving program performance measures, such as solution time or prediction ac-curacy. Local search approaches in particular have proven very effective for tuning algorithms. In se-quential local search, the use of predictive models has proven beneficial for obtaining good tuning re-sults. We study the use of non-parametric models in the context of population-based algorithm configu-rators. We introduce a new model designed specif-ically for the task of predicting high-performance regions in the parameter space. Moreover, we intro-duce the ideas of genetic engineering of offspring as well as sexual selection of parents. Numerical results show that model-based genetic algorithms significantly improve our ability to effectively con-figure algorithms automatically