Augmented compact genetic algorithm

Abstract

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 stochastically maintains the progress of convergence even after the probability (distribution) vector (PV) begins transition towards one of the solutions. Experimental results show that the proposed algorithm converges to a similar solution at a faster rate than the cGA. © Springer-Verlag Berlin Heidelberg 2004.