Surrogate‐based methods for black‐box optimization

Vu, Ky Khac; d'Ambrosio, Claudia; Hamadi, Youssef; Liberti, Leo

doi:10.1111/itor.12292

Cited by 123 publications

(64 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More practical co-evolutionary approaches, for example using only small numbers of populations for the outer search and the inner (uncertainty) search which share information between populations from generation to generation, or following several generations, require the application of additional simplifications and assumptions, see [CSZ09,MKA11]. One general area of research is the use of emulation to reduce the potential burden of computational run times and the number of modelfunction evaluations, see [VDYL16]. [ZZ10] use a surrogate-assisted evolutionary algorithm to tackle the inner search for black-box min max problems.…”

Section: Metaheuristic For Robust Optimisationmentioning

confidence: 99%

A largest empty hypersphere metaheuristic for robust optimisation with implementation uncertainty

Hughes

Goerigk

Wright

2019

Computers & Operations Research

View full text Add to dashboard Cite

We consider box-constrained robust optimisation problems with implementation uncertainty. In this setting, the solution that a decision maker wants to implement may become perturbed. The aim is to find a solution that optimises the worst possible performance over all possible perturbances.Previously, only few generic search methods have been developed for this setting. We introduce a new approach for a global search, based on placing a largest empty hypersphere. We do not assume any knowledge on the structure of the original objective function, making this approach also viable for simulation-optimisation settings. In computational experiments we demonstrate a strong performance of our approach in comparison with state-of-the-art methods, which makes it possible to solve even high-dimensional problems.

show abstract

Section: Metaheuristic For Robust Optimisationmentioning

confidence: 99%

A largest empty hypersphere metaheuristic for robust optimisation with implementation uncertainty

Hughes

Goerigk

Wright

2019

Computers & Operations Research

View full text Add to dashboard Cite

show abstract

“…71 Other tasks such as SA, analysis of variance, parameter screening, and model calibration can be efficiently conducted using the surrogate without worrying about the computational cost. 67,70,72,73 Building the surrogate can be done through a variety of methods based on classical machine learning algorithms, some of them that are related to this framework we highlight here. Polynomial chaos expansion (PCE) is one of the most common and earliest methods used to construct surrogate models, and it is expressed by [74][75][76]…”

Section: Machine Learning Deep Learning and Data Sciencementioning

confidence: 99%

Combining simulations and data with deep learning and uncertainty quantification for advanced energy modeling

Radaideh

Kozłowski

2019

Int J Energy Res

View full text Add to dashboard Cite

Summary A novel and modern framework for energy modeling is developed in this paper with a focus on nuclear energy modeling and simulation. The framework combines multiphysics simulations and real data, with validation by uncertainty quantification tasks and facilitation by machine and deep learning methods. The hybrid framework is built on the basis of a wide range of physical models, real data, mathematical and statistical methods, and artificial intelligence techniques. The framework is demonstrated in different applications, including quantifying uncertainties in computer simulations, multiphysics coupling, analysis of variance using machine learning surrogate models, deep learning of time series phenomena, and propagating parametric uncertainties of nuclear data. The applications demonstrated are oriented to nuclear engineering simulations, even though majority of the methods are applicable to other energy sources (eg, renewable). Efficient utilization of this framework is expected to yield a much better understanding of the physical phenomena analyzed as well as an improvement in the performance of the energy design/model under construction.

show abstract

“…Hence, kriging [38] is well adapted to construct such surrogate functions. Interested readers are invited to refer to [24,36,41] and references therein for a review on this subject. In our work, kriging consists in regarding the approximate predictionê as a special Gaussian stochastic process of spatial covariance given by…”

Section: Metamodeling and Adaptive Global Optimizationmentioning

confidence: 99%

Optimization of Fiber Orientation Model Parameters in the Presence of Flow-Fiber Coupling

Li¹,

Luyé²

2018

J. Compos. Sci.

View full text Add to dashboard Cite

In this paper, we propose a novel systematic procedure to minimize the discrepancy between the numerically predicted and the experimentally measured fiber orientation results on an injection-molded part. Fiber orientation model parameters are optimized simultaneously using Latin hypercube sampling and kriging-based adaptive surrogate modeling techniques. Via an adequate discrepancy measure, the optimized solution possesses correct skin–shell–core structure and global orientation evolution throughout the considered center-gated disk. Some non-trivial interaction between these parameters and flow-fiber coupling effects as well as their quantitative importance are illustrated. The parametric fine-tuning of orientation models mostly leads to a better agreement in the skin and shell regions, while the coupling effect via a fiber-dependent viscosity improves prediction in the core.

show abstract

Surrogate‐based methods for black‐box optimization

Cited by 123 publications

References 64 publications

A largest empty hypersphere metaheuristic for robust optimisation with implementation uncertainty

A largest empty hypersphere metaheuristic for robust optimisation with implementation uncertainty

Combining simulations and data with deep learning and uncertainty quantification for advanced energy modeling

Optimization of Fiber Orientation Model Parameters in the Presence of Flow-Fiber Coupling

Contact Info

Product

Resources

About