Genetic algorithm with multi-colony dynamic reproduction
Thanh-Do Tran, Gang-Gyoo Jin · 2010
In this paper, a new paradigm for genetic algorithms that makes use of the dynamic-system-model-based reproduction with the concept of multiple colonies is introduced. In the reproduction operator based on dynamic system model, information about location and fitness of the best individual within the current population is utilized to guide candidate solutions towards new locations closer to the best according to a dynamic-system-model-based guidance technique. This operator is designed to be favorable for exploitation. In order to overcome the loss of diversity and premature convergence issues, a multiple colonies version of the reproduction based on dynamic system model is introduced. Preliminary results from experimental study demonstrate the efficiency and effectiveness of the proposed algorithm.