Adaptive Information Granulation in Fitness Estimation for Evolutionary Optimization

作者:Tian, Jie; Zeng, Jianchao; Tan, Ying*; Sun, Chaoli
来源:Wireless Personal Communications, 2018, 103(1): 741-759.
DOI:10.1007/s11277-018-5474-2

摘要

Evolutionary algorithms ordinarily need to conduct lots of fitness evaluations, requiring big computational overhead particularly in complex optimization problems. This paper proposes an adaptive information granulation approach which inspired from on the granule computing, and then reduces the expensive original fitness evaluation by the aid of the fitness inheritance strategy based on the proposed adaptive information granulation approach. The proposed algorithm is compared with few fitness inheritance assisted evolutionary algorithm on both traditional benchmark problems with four different dimensions, the CEC 2013 functions and the CEC 2014 expensive optimization test problems with 30 dimensions. Experimental results show both high effectiveness and efficiency with better solutions than those compared algorithm within different finite budget of computation for different benchmark problems. Its advantages are further verified by a real-world light aircraft wing design problem.

全文