Evolutionary Computation as Survival Game for Individuals with Cooperative Strategies

Abstract

We previously proposed an evolutionary algorithm which views evolutionary computation as survival game for individuals. In the algorithm, each individual selects a place and a cycle time used for meeting other individuals at the beginning of each generation according to its own strategy. A place and a cycle time that an individual selects affect whether the individual can survive in next generation. However, in the previous study, all individuals took the same strategy that a place and a cycle time were not changed at every generation. In this paper, we propose two kinds of strategies. The strategies change a place and a cycle time of an individual at the beginning of each generation if the condition is met. The simulation results show that one of the two strategies yields longer lifetimes of individuals and obtain better fitness values of individuals for most test problems. The results also suggest that the evolutionary algorithm is suitable for investigation on the adaptation process of individuals in autonomous decentralized environments.