Abstract
The GPareto package for R provides multi-objective optimization algorithms for expensive black-box functions and an ensemble of dedicated uncertainty quantification methods. Popular methods such as efficient global optimization in the mono-objective case rely on Gaussian processes or kriging to build surrogate models. Driven by the prediction uncertainty given by these models, several infill criteria have also been proposed in a multi-objective setup to select new points sequentially and efficiently cope with severely limited evaluation budgets. They are implemented in the package, in addition with Pareto front estimation and uncertainty quantification visualization in the design and objective spaces. Finally, it attempts to fill the gap between expert use of the corresponding methods and user-friendliness, where many efforts have been put on providing graphical postprocessing, standard tuning and interactivity.
Highlights
Numerical modeling of complex systems is an essential process in fields as diverse as natural sciences, engineering, quality or economics
Many surrogate models are used in practice: polynomials, splines, support vector regression, radial basis functions, random forests or Gaussian processes (GP)
In Binois et al (2015a), an alternative relying on conditional simulations of Gaussian process models is detailed, which provides an estimate of the Pareto front and an associated measure of uncertainty
Summary
Numerical modeling of complex systems is an essential process in fields as diverse as natural sciences, engineering, quality or economics. Many surrogate models are used in practice: polynomials, splines, support vector regression, radial basis functions, random forests or Gaussian processes (GP). They may be integrated in various optimization strategies, see e.g., Wang and Shan (2007), Santana-Quintero, Montano, and Coello (2010), Tabatabaei, Hakanen, Hartikainen, Miettinen, and Sindhya (2015) and references therein. The GPareto package proposes Gaussian-Process based sequential strategies to solve multiobjective optimization (MOO) problems in a black-box, numerically expensive simulator context. It considers the case of models with multiple outputs, y(1)(x), .
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
