Kwang-Ju Institute of Science and Technology

An augmented compact genetic algorithm (acGA) is presented in this paper. It exhibits all the desirable characteristics of compact genetic algorithm (cGA). While the selection strategy of cGA is similar to (steady-state) tournament selection with replacement (TSR), the proposed algorithm employs a strategy akin to tournament selection without replacement (TS/R). The latter is a common feature of genetic algorithms (GAs) as it is perceived to be effective in keeping the selection noise as low as possible. The proposed algorithm maintains genetic diversity even after the probability vector begins transition towards one of the solutions. The genetic diversity leads to improved convergence performance for the most part. Computer simulations show that the proposed algorithm converges more quickly (to similar solution) than extant cGA.