||Research progress survey on interactive evolutionary computation
Pei, YanTakagi, Hideyuki
Journal of Ambient Intelligence and Humanized Computing
14 , 2018-05-23 , Springer Berlin Heidelberg
In this paper, we evaluate the acceleration performance of evolutionary computation by using estimated convergence points. The obtained convergence points can be obtained by the evolutionary computation algorithms with the characteristic that there is a one-to-one correspondence between a parent individual and its offspring. We can obtain multiple evolutionary paths or moving vectors for estimating convergence points for evolutionary computation algorithms acceleration. We use the estimation method and its four clustering improvement method for bimodal task to design a differential evolution with elite strategy. From the evaluation results, our proposed method can significantly enhance optimization performance of differential evolution. We also discuss the analyse the open topics and future opportunities on the estimation methods, some discoveries are investigated and analysed.