Spatial information sampling: another feedback mechanism of realising adaptive parameter control in meta-heuristic algorithms
Haichuan Yang, Sichen Tao, Zhiming Zhang, Zonghui Cai, Shangce Gao
Abstract
This paper innovatively proposes a spatial information sampling strategy to adaptively control the parameters of meta-heuristic algorithms (MHAs). The solutions' spatial distribution information in current iterations is used to control the parameters in the following iterations. An adaptive parameter control method requires obtaining information from the operation of MHAs and feeding it back to the adjustment of parameters. The mainstream information acquisition method is to record the changes to the solutions in the iterative process. In essence, the proposed feedback method, i.e., chaotic perceptron (CP), makes use of the temporal information arising from the change of solutions in MHAs. The wingsuit flying search algorithm and differential evolution are employed as case studies. Experimental results validate the effectiveness of the proposed strategy. The source code of CP can be found at https: //toyamaailab.github.io/.