Globally-biased BIRECT algorithm with local accelerators for expensive global optimization
作者：
Highlights：
• A new DIRECT-type derivative-free global optimization method is proposed.
• A new hybrid technique unifies a novel partition strategy with sampling on diagonals.
• Three local minimization strategies are neatly embedded into the global search.
• Comparison on more than 850 test problems shows advantages of the approach.
• Results on a real-life nonlinear regression model confirm the goodness of the algorithm.
摘要
•A new DIRECT-type derivative-free global optimization method is proposed.•A new hybrid technique unifies a novel partition strategy with sampling on diagonals.•Three local minimization strategies are neatly embedded into the global search.•Comparison on more than 850 test problems shows advantages of the approach.•Results on a real-life nonlinear regression model confirm the goodness of the algorithm.
论文关键词：Nonlinear global optimization,DIRECT-Type algorithms,BIRECT Algorithm,Hybrid optimization algorithms,Nonlinear regression
论文评审过程：Received 11 March 2019, Revised 23 August 2019, Accepted 23 October 2019, Available online 9 November 2019, Version of Record 19 November 2019.
论文官网地址：https://doi.org/10.1016/j.eswa.2019.113052
