A Sawtooth Growing Exploitation Framework for Memetic Algorithms
Xin Zhang, Yi Cai, Yue Sun, Lihua Sun, Zhou Wu · 2018
Memetic algorithms (MAs) refer to hybrid methods of global search and local search, which aims to systematically balance exploitation and exploration for solving an optimization problem. This paper attempts to create a sawtooth growing exploitation framework for MAs. Under the framework, a sawtooth-wave function is used to control ratios of global search and local search, and thus the exploitation is restricted to sawtooth growing patterns. An MA instance is implemented by combining modified differential evolution and neighborhood field algorithms, named as MDE-NF. Compared with several state-of-the-art algorithms, the MDE-NF algorithm shows promising performance on several benchmark functions.