Evolution strategies –A comprehensive introduction
Natural Computing: an international journal
Stochastic Global Optimization: Problem Classes and Solution Techniques
Journal of Global Optimization
Completely Derandomized Self-Adaptation in Evolution Strategies
Evolutionary Computation
Experimental Research in Evolutionary Computation: The New Experimentalism (Natural Computing Series)
Local meta-models for optimization using evolution strategies
PPSN'06 Proceedings of the 9th international conference on Parallel Problem Solving from Nature
Single- and multiobjective evolutionary optimization assisted by Gaussian random field metamodels
IEEE Transactions on Evolutionary Computation
Parameter tuning boosts performance of variation operators in multiobjective optimization
PPSN'10 Proceedings of the 11th international conference on Parallel problem solving from nature: Part I
When parameter tuning actually is parameter control
Proceedings of the 13th annual conference on Genetic and evolutionary computation
PSO based on surrogate modeling as meta-search to optimise evolutionary algorithms parameters
Proceedings of the 14th annual conference on Genetic and evolutionary computation
High-dimensional model-based optimization based on noisy evaluations of computer games
LION'12 Proceedings of the 6th international conference on Learning and Intelligent Optimization
| Hi-index | 0.00 |
The case-specific tuning of parameters of optimization metaheuristics like evolutionary algorithms almost always leads to significant improvements in performance. But if the evaluation of the objective function is computationally expensive, which is typically the case for real-worlds problems, an extensive parameter tuning phase on the original problem is prohibitive. Therefore we have developed another approach: Provided that a (computationally cheap) surrogate model is available that reflects the structural characteristics of the original problem then the parameter tuning can be run on the surrogate problem before using the best parameters thereby identified for the metaheuristic when optimizing the original problem. In this experimental study we aim to assess how many function evaluations on the original problem are necessary to build a surrogate model endowed with the characteristics of the original problem and to develop a methodology that measures to which extent such a matching has been achieved.