Abstrct
合作型協(xié)同進(jìn)化算法
Cooperative co-evolution (CC) is an effective framework that can be used to solve large-scale optimization problems. It typically divides a problem into components and uses one optimizer to solve the components in a round-robin fashion.
合作型協(xié)同進(jìn)化（CC）是一種有效的框架，可用于解決大規(guī)模優(yōu)化問題。 它通常將問題劃分為組件，并使用一個優(yōu)化器以循環(huán)方式解決組件。
問題
The relative contribution of each component to the overall fitness value may vary. Furthermore, using one optimizer may not be sufficient when solving a wide range of components with different characteristics.
每個組分對總體適合度值的相對貢獻(xiàn)可以變化。 此外，在解決具有不同特征的各種組件時，使用一個優(yōu)化器可能還不夠。
解決方案
We propose a novel CC framework which can select an appropriate optimizer to solve a component based on its contribution to the fitness improvement. In each evolutionary cycle, the candidate optimizer and component that make the greatest contribution to the fitness improvement are selected for evolving.
我們提出了一種新穎的CC框架，它可以根據(jù)其對適應(yīng)度提升的貢獻(xiàn)來選擇合適的優(yōu)化器來解決子問題。 在每個進(jìn)化周期中，選擇對適應(yīng)度提升做出最大貢獻(xiàn)的候選優(yōu)化器和子問題用于進(jìn)化。
實(shí)驗(yàn)結(jié)果
We evaluated the efficacy of the proposed CC with Optimizer Selection (CCOS) algorithm using large-scale benchmark problems. The numerical experiments showed that CCOS outperformed the CC model without optimizer selection ability. When compared against several other state-of-the-art algorithms, CCOS generated competitive solution quality.
我們使用大規(guī)?；鶞?zhǔn)問題評估了已提出的CC方法與優(yōu)化器選擇（CCOS）算法的效果。 數(shù)值實(shí)驗(yàn)表明，CCOS優(yōu)于沒有優(yōu)化選擇能力的CC模型。與其他幾種最先進(jìn)的算法相比，CCOS具有競爭力。
關(guān)鍵字
Large-scale optimization, cooperarive co-evolution, algorithm selection, algorithm hybridization, resources allocation
大規(guī)模優(yōu)化，合作型協(xié)同進(jìn)化，算法選擇，算法混合，資源分配
Conclusion
方法
In this paper, we have investigated how the use of alternative optimizers at different evolutionary stages impacted on the solution quality generated by the CC when used to solve LSGO problems. Instead of employing only one optimizer to solve all the components, we proposed an online optimizer selection framework to select the best optimizer from a portfolio for each component. At each evolutionary cycle, the component and optimizer pair that previously contributed the most to the overall fitness improvement was selected for evolving.
在本文中，我們研究了在用于解決LSGO問題時，在不同演化階段使用替代優(yōu)化器如何影響CC產(chǎn)生的解的質(zhì)量。 我們提出了一個在線優(yōu)化器選擇框架，針對每個子問題，從集合中選擇最佳優(yōu)化器，而不是僅使用一個優(yōu)化器來解決所有子問題。 在每個進(jìn)化周期中，選擇先前對整體適應(yīng)性改善貢獻(xiàn)最大的子問題和優(yōu)化器對進(jìn)行演化。
實(shí)驗(yàn)
We experimentally demonstrated that the proposed CCOS algorithm was successful in selecting the best optimizer when solving the CEC’2010 benchmark problems. Significantly, CCOS could potentially generate statistically better solution quality than the default CC algorithm with no optimizer selection ability. When compared against several other state-of-the-art algorithms, CCOS also achieved competitive results.
我們通過實(shí)驗(yàn)證明，在解決CEC'2010基準(zhǔn)問題時，所提出的CCOS算法成功地選擇了最佳優(yōu)化器。 值得注意的是，CCOS可能比沒有優(yōu)化器選擇能力的默認(rèn)CC算法產(chǎn)生統(tǒng)計(jì)上更好的解決方案質(zhì)量。 與其他幾種最先進(jìn)的算法相比，CCOS也取得了有競爭力的成果。
