Theoretical Analysis of Simple Evolution Strategies in Quickly Changing Environments

摘要

Evolutionary algorithms applied to dynamic optimization problems has become a promising research area. So far, all papers in the area have assumed that the environment changes only between generations. In this paper, we take a first look at possibilities to handle a change during a generation. For that purpose, we derive an analytical model for a (1,2) evolution strategy and show that sometimes it is better to ignore the environmental change until the end of the generation, than to evaluate each individual with the most up-to-date fitness function.