摘要: In 1999, Harik and Lobo [1] proposed a technique that automates the choice of the population size parameter of a genetic algorithm by simulating a race among an unbounded number of populations of exponentially increasing sizes, with the smaller sized populations being allowed more execution time than larger ones. The proposed method uses a m-ary counter in its implementation.In this thesis, the student will explore the design of an alternative implementation for the same idea by using a randomized algorithm. The new algorithm is likely to be simpler to implement and have the same behavior as the original one. It is expected that the student implements and makes a comparison of both approaches.