Completely Derandomized Self-Adaptation in Evolution Strategies
Evolutionary Computation
The Journal of Machine Learning Research
On the generality of parameter tuning in evolutionary planning
Proceedings of the 12th annual conference on Genetic and evolutionary computation
An evaluation of off-line calibration techniques for evolutionary algorithms
Proceedings of the 12th annual conference on Genetic and evolutionary computation
On the benefit of sub-optimality within the divide-and-evolve scheme
EvoCOP'10 Proceedings of the 10th European conference on Evolutionary Computation in Combinatorial Optimization
Hi-index | 0.00 |
Learn-and-Optimize (LaO) is a generic surrogate based method for parameter tuning combining learning and optimization. In this paper LaO is used to tune Divide-and-Evolve (DaE), an Evolutionary Algorithm for AI Planning. The LaO framework makes it possible to learn the relation between some features describing a given instance and the optimal parameters for this instance, thus it enables to extrapolate to unknown instances in the same domain. Moreover, the learned model is used as a surrogate-model to accelerate the search for the optimal parameters. The proposed implementation of LaO uses an Artificial Neural Network for learning the mapping between features and optimal parameters, and the Covariance Matrix Adaptation Evolution Strategy for optimization. Results demonstrate that LaO is capable of improving the quality of the DaE results even with only a few iterations. The main limitation of the DaE case-study is the limited learning time and amount of meaningful features that are available. However, it is demonstrated that the learned model is capable of generalization in the domain to unknown instances.