Cultural algorithms: knowledge learning in dynamic environments

Our previous work on real-valued function optimization problems had shown that cultural learning emerged as the result of meta-level interaction or swarming of knowledge sources, "knowledge swarms" in the belief space. These meta-level swarms induced the swarming of individuals in the population space, "cultural swarms". The interaction of these knowledge sources with the population swarms produced three emergent phases of problem solving. This reflected an algorithmic process that emerged from the interaction of the knowledge sources. We investigate the extent to which these emergent phenomena are also visible within dynamic environments. We motivate the discussion in terms of a simulation model of a reversible switching surface. We demonstrate how we can program such changes in surface structure using knowledge source interaction in cultural algorithms.