This paper gives a comprehensive comparison between real population based tournament selection and virtual population based tournament selection both in theory and in experiment. We claim that virtual population based EDA is able to obtain a comparative or even better solution when compared with the one captured by real population based EDA. EDA using virtual population based tournament selection does not store the whole population in the memory. Therefore, less memory is required when compared with the one using real population based tournament selection. Another advantage of EDA using virtual population based tournament selection is higher selection intensity can be achieved, therefore less fitness evaluations are needed to converge.
展开▼
