Batch Sequential Adaptive Designs for Global Optimization

Ning, Jianhui; Xiao, Yao; Xiong, Zikang

doi:10.48550/arxiv.2010.10698

Cited by 1 publication

(3 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this subsection, we will answer this question with numerical results. We compared the SCO method with some existing batch methods, including the q-EI method with kriging believer (KB) and constant liar minimum (CL-min) as heuristic strategies (Ginsbourger et al, 2010), multi-point sampling based on kriging (MPSK) method (Cai et al, 2017) and accelerated EGO (aEGO) method (Ning et al, 2020).…”

Section: Efficiency Of Batch Bayesian Optimizationmentioning

confidence: 99%

“…We compare SCO method with some existing batch methods, including q-EI method with "KB" and "CL-min" as heuristic strategies (Ginsbourger et al, 2010), MPSK method (Cai et al, 2017) and aEGO (Ning et al, 2020). Note that all these methods build a Gauss process model as a surrogate.…”

Section: Efficiency Of Batch Bayesian Optimizationmentioning

confidence: 99%

“…Cai et al (2017) and Wang et al (2004) use the mode-pursuing sampling method (Fu and Wang, 2002) to obtain the sequential design. Ning et al (2020) use the samplingimportance-resampling method to accelerate EGO. However, sampling methods are arbitrary and uncertain, which is not appropriate in an expensive experiment.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Less Uncertain Sampling-Based Method of Batch Bayesian Optimization

Jia¹,

Duan²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper presents a method called sampling-computation-optimization (SCO) to design batch Bayesian optimization. SCO does not construct new high-dimensional acquisition functions but samples from the existing one-site acquisition function to obtain several candidate samples. To reduce the uncertainty of the sampling, the general discrepancy is computed to compare these samples. Finally, the genetic algorithm and switch algorithm are used to optimize the design. Several strategies are used to reduce the computational burden in the SCO. From the numerical results, the SCO designs were less uncertain than those of other sampling-based methods. As for application in batch Bayesian optimization, SCO can find a better solution when compared with other batch methods in the same dimension and batch size. In addition, it is also flexible and can be adapted to different one-site methods. Finally, a complex experimental case is given to illustrate the application value and scenario of SCO method.

show abstract

Section: Efficiency Of Batch Bayesian Optimizationmentioning

confidence: 99%