Category Archives: Surrogate Models

Meeting with Partners from DLR (Deutsche Zentrum für Luft- und Raumfahrt) @TH_Koeln @utopiae_network

Dr. Stefan Görtz (Deutschen Zentrum für Luft- und Raumfahrt (DLR), Institut für Aerodynamik und Strömungstechnik) and Christian Sabater (Early Stage Researcher in the UTOPIAE Project) visited the SPOTSeven Lab at TH Köln in Gummersbach. Here are some impressions from our inspiring discussions about uncertainty, reliability, optimization, surrogates, etc.

This slideshow requires JavaScript.

Two Full Papers Accepted @GECCO2018

Our papers
“Surrogate assisted optimization of particle reinforced metal matrix composites” (L. Gentile, M. Zaefferer, D. Giugliano, H. Chen, T. Bartz-Beielstein)
and
“Comparison of Parallel Surrogate-Assisted Optimization Approaches” (F. Rehbach, M. Zaefferer, J. Stork, T. Bartz-Beielstein)
have been accepted as full papers. 100% of our papers, i.e., two out of two, were successful. We are looking forward to giving the presentations in Kyoto!

New paper: #Optimization via multimodel #simulation published in “Structural and Multidisciplinary Optimization”

An online version of the paper “Optimization via multimodel simulation” (https://doi.org/10.1007/s00158-018-1934-2) written by Thomas Bartz-Beielstein, Martin Zaefferer, and Quoc Cuong Pham was published today. This research paper was published in the journal “Structural and Multidisciplinary Optimization“.

Abstract:
“Increasing computational power and the availability of 3D printers provide new tools for the combination of modeling and experimentation. Several simulation tools can be run independently and in parallel, e.g., long running computational fluid dynamics simulations can be accompanied by experiments with 3D printers. Furthermore, results from analytical and data-driven models can be incorporated. However, there are fundamental differences between these modeling approaches: some models, e.g., analytical models, use domain knowledge, whereas data-driven models do not require any information about the underlying processes. At the same time, data-driven models require input and output data, but analytical models do not. The optimization via multimodel simulation (OMMS) approach, which is able to combine results from these different models, is introduced in this paper. We believe that OMMS improves the robustness of the optimization, accelerates the optimization-via-simulation process, and provides a unified approach. Using cyclonic dust separators as a real-world simulation problem, the feasibility of this approach is demonstrated and a proof-of-concept is presented. Cyclones are popular devices used to filter dust from the emitted flue gasses. They are applied as pre-filters in many industrial processes including energy production and grain processing facilities. Pros and cons of this multimodel optimization approach are discussed and experiences from experiments are presented.”

Keywords
Combined simulation Multimodeling Simulation-based optimization Metamodel Multi-fidelity optimization Stacking Response surface methodology 3D printing Computational fluid dynamics

Cite this article as
Bartz-Beielstein, T., Zaefferer, M. & Pham, Q.C. Struct Multidisc Optim (2018). https://doi.org/10.1007/s00158-018-1934-2

Publisher Name
Springer Berlin Heidelberg
Print ISSN1615-147X
Online ISSN1615-1488

In a Nutshell: Sequential Parameter #Optimization is also on @arxiv #rstats @utopiae_network

The paper “In a Nutshell: Sequential Parameter Optimization” has been assigned the permanent arXiv identifier 1712.04076 and is available at:
http://arxiv.org/abs/1712.04076

arXiv:1712.04076
Date: Tue, 12 Dec 2017 00:03:45 GMT   (2255kb,D)
Title: In a Nutshell: Sequential Parameter Optimization
Authors: Thomas Bartz-Beielstein, Lorenzo Gentile, Martin Zaefferer
Categories: cs.MS cs.AI math.OC
Comments: Version 12/2017
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Abstract:
The performance of optimization algorithms relies crucially on their
parameterizations. Finding good parameter settings is called algorithm tuning.
Using a simple simulated annealing algorithm, we will demonstrate how
optimization algorithms can be tuned using the sequential parameter
optimization toolbox (SPOT). SPOT provides several tools for automated and
interactive tuning. The underling concepts of the SPOT approach are explained.
This includes key techniques such as exploratory fitness landscape analysis and
response surface methodology. Many examples illustrate how SPOT can be used for
understanding the performance of algorithms and gaining insight into
algorithm’s behavior. Furthermore, we demonstrate how SPOT can be used as an
optimizer and how a sophisticated ensemble approach is able to combine several
meta models via stacking.

“Metamodel-based optimization of hot rolling processes in the metal industry” – Full-text view-only version available

The article “Metamodel-based optimization of hot rolling processes in the metal industry” was published in The International Journal of Advanced Manufacturing. A full-text view-only version of this paper is available. All readers of this article via the shared link will also be able to use Enhanced PDF features such as annotation tools, one-click supplements, citation file exports and article metrics. Here is the link:
http://rdcu.be/vvws

Free Paper: In a Nutshell – Sequential Parameter #Optimization @utopia_network #TH_Koeln #rstats

We proudly present the first result of our cooperation in the UTOPIAE network:
Lorenzo Gentile, Martin Zaefferer, and Thomas Bartz-Beielstein published a paper about sequential parameter optimization. The paper is based on the talk given by Thomas Bartz-Beielstein during the UTOPIAE Network Training School at University of Strathclyde, Glasgow, earlier this year.

The paper is entitled “In a Nutshell: Sequential Parameter Optimization”. It gives a short introduction to surrogate model based optimization, which will be applied in the UTOPIAE project. Many examples illustrate the usefulness of the SPOT approach.

The paper can be downloaded here.

The abstract reads as follows:
The performance of optimization algorithms relies crucially on their parameterizations. Finding good parameter settings is called algorithm tuning. Using a simple simulated annealing algorithm, we will demonstrate how optimization algorithms can be tuned using the sequential parameter opti- mization toolbox (SPOT). SPOT provides several tools for automated and interactive tuning. The underling concepts of the SPOT approach are explained. This includes key techniques such as exploratory fitness landscape analysis and response surface methodology. Many examples illustrate how SPOT can be used for understanding the performance of algorithms and gaining insight into algorithm’s behavior. Furthermore, we demonstrate how SPOT can be used as an optimizer and how a sophisticated ensemble approach is able to combine several meta models via stacking.

Sequential Parameter #Optimization Toolbox 2.0.2 is on CRAN #RSTATS


Version 2.0.2 of SPOT, the Sequential Parameter Optimization Toolbox, is available on CRAN:
https://cran.r-project.org/web/packages/SPOT/index.html
SPOT provides a set of tools for model based optimization and tuning of algorithms. It includes surrogate models, optimizers and design of experiment approaches. The main interface is spot, which uses sequentially updated surrogate models for the purpose of efficient optimization. The main goal is to ease the burden of objective function evaluations, when a single evaluation requires a significant amount of resources.
Version 2 of the SPOT package is a complete redesign and rewrite of the original R package. Most function interfaces were redesigned to give a more streamlined usage experience. At the same time, modular and transparent code structures allow for increased extensibility. In addition, some new developments were added to the SPOT package. A Kriging model implementation, based on earlier Matlab code by Forrester et al. (Forrester, Sobester, and Keane 2008), has been extended to allow for the usage of categorical inputs. Additionally, it is now possible to use stacking for the construction of ensemble learners (Bartz-Beielstein and Zaefferer 2017). This allows for the creation of models with a far higher predictive performance, by combining the strengths of different modeling approaches.

References