A Spark-based Gaussian Bare-bones Cuckoo Search with Dynamic Parameter Selection

Zhihui He,Hu Peng,Changshou Deng,Yucheng Tan,Zhijian Wu,Shuangke Wu

DOI: https://doi.org/10.1109/cec.2019.8790040

2019-01-01

Abstract:Cuckoo search algorithm (CS), as a new heuristic algorithm, has been paid increasing attention and studied by many scholars because of its efficient performance. However, premature convergence is a defect of CS. Recently, some heuristic algorithms have been successfully applied to some high performance computing frameworks, effectively overcome the premature convergence problem of the algorithm, which provides us with a new idea to enhance CS. Therefore, a novel CS variant, called Spark-based gaussian bare-bones cuckoo search with dynamic parameter selection (SparkGDCS), which combines a novel CS variant with the efficient Spark framework, is proposed in this paper. In SparkGDCS, GDCS is a new variant which combines Gaussian bare-bones strategy and dynamic parameter selection, whose purpose is to enhance the search ability of CS. Finally, by testing the benchmark functions proposed by the 2010 and 2013 IEEE Congress on Evolutionary Computation special session (CEC 2010 and CEC 2013), comprehensive experiments prove the effectiveness of SparkGDCS.