Evolutionary Algorithms with Adaptive Levy Mutations

An evolutionary programming algorithm with adaptivemutation operators based on L evy prob ability distribution is studied L evy stable distri bution has an in nite second moment Because of this L evy mutation is more likely to generate an o spring that is farther away from its parent than Gaussian mutation which is often used in evolu tionary algorithms Such likelihood depends on a parameter in the distribution Based on this we propose an adaptive L evy mutation in which four di erent candidate o spring are generated by each parent according to and and the best one is chosen as the o spring for the next generation The proposed algorithm was applied to several multivariate function opti mization problems We showed empirically that the performance of the proposed algorithm was better than that of classical evolutionary algo rithms using Gaussian mutation