A study on metamodeling techniques, ensembles, and multi-surrogates in evolutionary computation

Lim, Dudy; Ong, Yew-Soon; Jin, Yaochu; Sendhoff, Bernhard

doi:10.1145/1276958.1277203

Cited by 72 publications

(26 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To reduce the needed number of expensive function evaluations, surrogate-assisted evolutionary algorithms (SAEAs) [22], [8], [21] can be used, where the main idea is to use one [23], [24] or multiple surrogates [25], [26], [27] to approximate the expensive function evaluation globally or locally [28], [29], [30]. Note that an implicit assumption here is that the computational cost for constructing and using the surrogates is much less than that for fitness evaluations using the original expensive function.…”

Section: B Surrogate Models and Surrogate Managementmentioning

confidence: 99%

Data-Driven Surrogate-Assisted Multiobjective Evolutionary Optimization of a Trauma System

Wang

Jin

Jansen

2016

IEEE Trans. Evol. Computat.

Self Cite

188

100

View full text Add to dashboard Cite

Abstract-Most existing work on evolutionary optimization assumes that there are analytic functions for evaluating the objectives and constraints. In the real-world, however, the objective or constraint values of many optimization problems can be evaluated solely based on data and solving such optimization problems is often known as data-driven optimization. In this paper, we divide data-driven optimization problems into two categories, i.e., off-line and on-line data-driven optimization, and discuss the main challenges involved therein. An evolutionary algorithm is then presented to optimize the design of a trauma system, which is a typical off-line data-driven multi-objective optimization problem, where the objectives and constraints can be evaluated using incidents only. As each single function evaluation involves large amount of patient data, we develop a multi-fidelity surrogate management strategy to reduce the computation time of the evolutionary optimization. The main idea is to adaptively tune the approximation fidelity by clustering the original data into different numbers of clusters and a regression model is constructed to estimate the required minimum fidelity. Experimental results show that the proposed algorithm is able to save up to 90% of computation time without much sacrifice of the solution quality.

show abstract

Section: B Surrogate Models and Surrogate Managementmentioning

confidence: 99%

Data-Driven Surrogate-Assisted Multiobjective Evolutionary Optimization of a Trauma System

Wang

Jin

Jansen

2016

IEEE Trans. Evol. Computat.

Self Cite

188

100

View full text Add to dashboard Cite

show abstract

“…rmse of the 20 clustered localized models for variable levels of residual error tolerances, while taking the original airfoil analysis code as the reference model (i.e., with a fidelity of 1) case, this implies the correlation between the predicted fitness values is of utmost importance. For greater details, the reader is referred to [16][17][18], where it was shown that the correlation metric represents a more useful metric for estimating the quality of an approximation model over one that is based on absolute error.…”

Section: Multi-fidelity Models In the Aerodynamic Airfoil Design Problemmentioning

confidence: 99%

“…Particularly, local models are used in favor of global models since constructing accurate global models is fundamentally flawed due to the curse of dimensionality [18][19][20][21]. Further, this allows more precise estimation on the unique characteristics of the problem landscapes, thus leading to the prediction on the appropriate level of localized model fidelity over the use of the original computationally expensive model.…”

Section: : End Whilementioning

confidence: 99%

Evolutionary Optimization with Dynamic Fidelity Computational Models

Lim

Ong

Jin

et al. 2008

Advanced Intelligent Computing Theories and Applications. With Aspects of Artificial Intelligence

Self Cite

View full text Add to dashboard Cite

Abstract. In optimization, it is now a common practice to use lower fidelity computational models in place of the original model when dealing with problems with computationally expensive objective functions. In this paper, we present a study on evolutionary optimization with dynamic fidelity computational models capable of acclimatizing to localized complexity, for enhancing design search efficiency. In particular, we propose an evolutionary framework for model fidelity control that decides, at runtime, the appropriate fidelity level of the computational model, which is deemed to be computationally less expensive, to be used in place of the exact analysis code as the search progresses. Empirical study on an aerodynamic airfoil design problem based on a Memetic Algorithm with Dynamic Fidelity Model (MA-DFM) demonstrates that improved quality solution and efficiency are obtained over existing evolutionary schemes.

show abstract

“…Currently, the common fitness estimation methods include the fitness inheritance and the application of surrogate model [11][12][13][14][15][16][17]. However, which method will perform better in fitness estimation?…”

Section: Introductionmentioning

confidence: 99%

Efficient Query for Historical Data in Evolutionary Algorithm

Tian¹,

Pan²,

Huang³

2017

Proceedings of the 2017 International Conference on Electronic Industry and Automation (EIA 2017)

View full text Add to dashboard Cite

Abstract-For the time-consuming problem in calculating the fitness value, this paper proposes a hash bucket with precision mechanism for a quick query of the data in the neighborhood of a particle. In order to establish a balance between the calculation accuracy and utility, it uses the hash tables with precision mechanism to solve the problem in the storage and query of historical calculation data so that the neighborhood of a to-be-evaluated individual can be determined more accurately to reduce the error in estimating the fitness value. Moreover, it uses the typical reference functions to separately test the effectiveness and accuracy of the algorithms based on the values obtained in different dimensions. The test result proves that compared with the other algorithms described in this paper, our algorithm can provide a better solution in the context of the same number of times for fitness calculation.

show abstract

A study on metamodeling techniques, ensembles, and multi-surrogates in evolutionary computation

Cited by 72 publications

References 23 publications

Data-Driven Surrogate-Assisted Multiobjective Evolutionary Optimization of a Trauma System

Data-Driven Surrogate-Assisted Multiobjective Evolutionary Optimization of a Trauma System

Evolutionary Optimization with Dynamic Fidelity Computational Models

Efficient Query for Historical Data in Evolutionary Algorithm

Contact Info

Product

Resources

About