使用離散粒子群演算法尋找最佳無摺疊均勻實驗設計

Using Discrete Particle Swarm Optimization to Find Optimal Non-collapsing and Space-filling Experimental Designs

李其澔(Chi-Hao Li)

指導教授 ： 王偉仲

國立臺灣大學/理學院/數學研究所/碩士(2014年)

https://doi.org/10.6342/NTU.2014.01659

摘要

無資料

關鍵字

電腦實驗 ； 試驗設計 ； 均勻性質 ； 無摺疊性質 ； 離散粒子群演算法

並列摘要

Uniformity of experimental designs is an important issue in computer experiments recent years. To reduce the cost of handling experiment, we need to find usable designs effectively and effeciently. A design with good space-filling and non-collapsing properties may help us get the most information under some specific cost. Since a lot amount of real problems require grid discretization, based on the framework of the discrete particle swarm optimization (DPSO), we try several strategies and propose several applied methods to discuss the multi-objective issue, and will illustrate it by handling experiments on several regular and irregular feasible domains by some DPSO-based algorithms.

並列關鍵字

Computer Experiment ； Space-filling ； Uniform Experimental Designs ； Non-collapsing ； Discrete Particle Swarm Optimization

參考文獻

[1] Ray-Bing Chen, Dai-Ni Hsieh, Ying Hung, and Weichung

Wang. Optimization latin gypercube designs by particle swarm.

Staticstics and Computing, pages 1-14, 2012.

[2] Ray-Bing Chen, Yen-Wen Hsu, and Weichung Wang. Central

[3] Karel Crombecq, Eric Laermans, and Tom Dhaene. Efficient