In this study, we propose a new paradigm of evolutionary computation named "Eco-Evolution" for large- scale and complex real problems. The aim of this paradigm is to maintain the diversity of a population and to achieve convergence towards the global optimum and local optima in the evolutionary process. The idea of the paradigm is inspired from Tierra as an evolutionary ecological system that has widespread diversity. A concept of diversity in terms of applying evolutionary computation to real problems is defined. Several ideas are introduced for the implementation of this feature. They include individual lifetime and individual resource consumption based on fitness, and local crossover for clustering individuals in a population. In this study, Eco-Evolution is confirmed to have the feature of defined diversity.
展开▼