Genetic Algorithm with Adaptive Population

摘要

A new type of GA named VPGA with adaptive population size is presented. By constructing a function of the probabilty for VPGA to reach the optimal solution, we prove that if the population size is chosen according to a proper function, VPGA has higher probability to obtain optimal solution with less cost than that of other modified GAs such as modGA and finGA. Enlightened by the law of human population evolution in demography, we also suggest a logistic model of population size. Our theoretical and experimental results show that VPGA can obtain better solution with less computational cost.