A Dual-Population Genetic Algorithm for Balanced Exploration and Exploitation

T. Park and K.R. Ryu (Korea)


Genetic Algorithm, Optimization, balance of exploration and exploitation


This paper proposes a novel evolutionary algorithm named dual-population genetic algorithm (DPGA). The proposed algorithm has two distinct populations with different evolutionary objectives: the prospect population and the preserver population. The prospect population is similar to the population of a traditional genetic algorithm; however, it only consists of the winner of a local competition after a crossover. Although this local competition helps to find a local peak in a search space, it also causes the population to easily lose the diversity. The preserver population helps to maintain the diversity by preserving some of those chromosomes which are the losers of the local competition. Experiments with one max problem, deceptive function, royal road function, and TSP have shown that DPGA outperforms traditional genetic algorithm.

Important Links:

Go Back