Surrogate Modeling

Over the years, the accuracy of engineering simulation software has improved significantly allowing simulation of systems at a finer level of detail. This evolution opens up usage of simulations for increasingly complex problems, but also increases the associated computational cost tremendously.

Surrogate modeling is an interdisciplinary research field that uses data-efficient machine learning to expedite the analysis and optimization of high-fidelity simulations what otherwise may take weeks or months. Surrogate models are fast-running approximations of complex time-consuming computer simulations. They are also known as response surface models (RSM), metamodels, proxy models or emulators.

Surrogate modeling bridges the gap between the numerical or experimental, and the analytical. Surrogate models are used for parametric studies, optimization, design-space exploration, visualization, prototyping, uncertainty quantification and sensitivity analysis.

Staff

Tom Dhaene, Dirk Deschrijver

Researchers

Joachim van der Herten, Domenico Spina, Ivo Couckuyt, Tom Van Steenkiste, Nicolas Knudde,

Projects

• ICON FORWARD, “Creating robust and reliable wireless networks for harsh industrial environments”
• ICON SENCOM: “Smart energy consumption in manufacturing”
• TETRA NEATH: “NEATH - Near-field meettechnieken voor het efficiënt oplossen van emc emissie problemen”,
• ICON QOCON: “Quality Of service in COgnitive Networks”

Key publications

• Gorissen, Dirk, Ivo Couckuyt, Piet Demeester, Tom Dhaene, and Karel Crombecq. 2010. “A Surrogate Modeling and Adaptive Sampling Toolbox for Computer Based Design.” Journal of Machine Learning Research 11: 2051–2055.
• Couckuyt, Ivo, Tom Dhaene, and Piet Demeester. 2014. “ooDACE Toolbox: a Flexible Object-oriented Kriging Implementation.” Journal of Machine Learning Research 15: 3183–3186.
• Couckuyt, Ivo, Dirk Deschrijver, and Tom Dhaene. 2014. “Fast Calculation of Multiobjective Probability of Improvement and Expected Improvement Criteria for Pareto Optimization.” Journal of Global Optimization 60 (3): 575–594.
• van der Herten, Joachim, Ivo Couckuyt, Dirk Deschrijver, and Tom Dhaene. 2015. “A Fuzzy Hybrid Sequential Design Strategy for Global Surrogate Modeling of High-dimensional Computer Experiments.” Siam Journal on Scientific Computing 37 (2): A1020–A1039.   
• Adaptive classification under computational budget constraints using sequential data gathering. J van der Herten, I Couckuyt, D Deschrijver, T Dhaene. Advances in Engineering Software 99, 137-146