Combining the strengths of the Bayesian optimization algorithm and adaptive evolution strategies
Martin Pelikán, David E. Goldberg, Shigeyoshi Tsutsui · 2002
This paper proposes a method that combines competent genetic algorithms working in dis-crete domains with adaptive evolution strate-gies working in continuous domains. We use discretization to transform solution between the two domains. The results of our exper-iments with the Bayesian optimization al-gorithm as a discrete optimizer and -self-adaptive mutation of evolution strategies as a continuous optimizer combined using k-means clustering suggest that the algorithm scales up well on all tested problems. The proposed method can be used to ll the gap between other optimization methods working in continuous and discrete domains and allow their hybridization. 1