Multidimensional sequential sampling for NURBs-based metamodel development

Turner, Cameron J.; Crawford, Richard H.; Campbell, Matthew I.

doi:10.1007/s00366-006-0051-9

Cited by 28 publications

(21 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For space-filling sampling as proposed by Turner et al a normalized proximity function is constructed as a tensor product of parabolic spans between the nearest control points of the NURBS surface (Turner et al, 2007). If the control points coincide with the data points, then this method should produce very similar results as the proposed approach.…”

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

“…The literature on sequential sampling is full of adaptive space-filling sampling methods that work in conjunction with metamodels (for example: Koehler and Owen, 1996;Jones, 2001;Jin et al, 2002;Sasena, 2002;Turner et al, 2007). The proposed method does not rely upon information obtained from metamodels and the associated costs of training them, and hence this review does not cover these and derivative methods any further.…”

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

“…Sasena (2002) proposes a strategy that switches between exploitation and exploration, and Turner et al (2007) propose a multiobjective formulation that blends four different sampling goals. One of the strengths of the proposed adaptive search methods is the ability to switch between exploratory and exploitive search strategies.…”

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

“…Maximin (Johnson et al, 1990), qualitative and quantitative sequential sampling (Q2S2) (Rai and Campbell, 2008), the proximity-based method proposed for use in conjunction with NURBs-based metamodels (Turner et al, 2007), and minimal Kullback-Liebler information designs (Jourdan and Franco, 2009). These methods are reviewed next.…”

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

See 3 more Smart Citations

Bayesian Network Classifiers for Set-Based Collaborative Design

Shahan

Seepersad

2010

Volume 1: 36th Design Automation Conference, Parts a and B

View full text Add to dashboard Cite

Complex design problems are typically decomposed into smaller design problems that are solved by domain-specific experts who must then coordinate their solutions into a satisfactory system-wide solution. In set-based collaborative design, collaborating engineers coordinate themselves by communicating multiple design alternatives at each step of the design process. The goal in set-based collaborative design is to spend additional resources exploring multiple options in the early stages of the design process, in exchange for less iteration in the latter stages, when iterative rework tends to be most expensive. Several methods have been proposed for representing sets of designs, including intervals, surrogate models, fuzzy membership functions, and probability distributions. In this paper, we introduce the use of Bayesian networks for capturing sets of promising designs, thereby classifying the design space into satisfactory and unsatisfactory regions. The method is compared to intervals in terms of its capacity to accurately classify satisfactory design regions as a function of the number of available data points. A simplified, multilevel design problem for an unmanned aerial vehicle is presented as the motivating example.

show abstract

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

Section: Latin Hypercube Hammersleymentioning

confidence: 99%

See 2 more Smart Citations

Bayesian Network Classifiers for Set-Based Collaborative Design

Shahan

Seepersad

2010

Volume 1: 36th Design Automation Conference, Parts a and B

View full text Add to dashboard Cite

show abstract

“…Additional assumptions are that f is expensive to compute. Thus the number of function evaluations f (X) needs to be minimized and data points must be selected iteratively, at points where the information gain will be the greatest [96]. Mathematically this means defining a sampling function…”

Section: Global Surrogate Modelingmentioning

confidence: 99%

Multiobjective global surrogate modeling, dealing with the 5-percent problem

Gorissen

Couckuyt

Laermans

et al. 2009

Engineering with Computers

View full text Add to dashboard Cite

When dealing with computationally expensive simulation codes or process measurement data, surrogate modeling methods are firmly established as facilitators for design space exploration, sensitivity analysis, visualization, prototyping and optimization. Typically the model parameter (=hy-perparameter) optimization problem as part of global surrogate modeling is formulated in a single objective way. Models are generated according to a single objective (accuracy). However, this requires an engineer to determine a single accuracy target and measure upfront, which is hard to do if the behavior of the response is unknown. Likewise, the different outputs of a multi-output system are typically modeled separately by independent models. Again, a multiobjective approach would benefit the domain expert by giving information about output correlation and enabling automatic model type selection for each output dynamically. With this paper the authors attempt to increase awareness of the subtleties involved and discuss a number of solutions and applications. In particular we present a multiobjective framework for automatic global surrogate model generation to help tackle both problems and that is applicable in both the static and sequential design (adaptive sampling) case.

show abstract

Automatic Approximation of Expensive Functions with Active Learning

Gorissen

Crombecq

Couckuyt

et al. 2009

Studies in Computational Intelligence

View full text Add to dashboard Cite

Abstract. The use of computer simulations has become a viable alternative for real-life controlled experiments. Due to the computational cost of these high fidelity simulations, surrogate models are often employed as an approximation for the original simulator. Because of their compact formulation and negligible evaluation time, global surrogate models are very useful tools for exploring the design space, optimization, visualization and sensitivity analysis. Additionally, multiple surrogate models can be chained together to model large scale systems where the direct use of the original expensive simulators would be too cumbersome. Many surrogate model types, such as neural networks, support vector machines and rational models have been proposed, and many more techniques have been developed to minimize the number of expensive simulations required to train a sufficiently accurate surrogate model. In this chapter, we present a fully automated and integrated global surrogate modeling methodology for regression modeling and active learning that readily enables the adoption of advanced global surrogate modeling methods by application scientists. The work brings together insights from distributed systems, artificial intelligence, and modeling & simulation, and has applications in a very wide range of fields. The merits of this approach are illustrated with several examples, and several surrogate model types.

show abstract

Multidimensional sequential sampling for NURBs-based metamodel development

Cited by 28 publications

References 27 publications

Bayesian Network Classifiers for Set-Based Collaborative Design

Bayesian Network Classifiers for Set-Based Collaborative Design

Multiobjective global surrogate modeling, dealing with the 5-percent problem

Automatic Approximation of Expensive Functions with Active Learning

Contact Info

Product

Resources

About