Cooperative Co-Evolution With Formula Based Grouping and CMA for Large Scale Optimization

Cooperative co-evolution framework is widely used in large scale optimization problems. Usually, the large scale problem is divided into smaller sub groups using black-box decomposition methods based on variable interactions. However these black-box decomposition methods have limitations in finding correct variable interactions. In this paper, a white-box decomposition method named formula based grouping(FBG) is adopted and further improved. Also, we extend the covariance matrix adaptation to work with FBG under the cooperative co-evolution framework. Based on it, a new evolutionary algorithm is proposed for handling large scale optimization problems. The numerical experiments on CEC' 2013 benchmark suit shows the efficiency of the proposed algorithm.