M. Eng. Martin Zaefferer
Martin Zaefferer was awarded with the Ferchau-Förderpreis and the FESTO-Förderpreis. His master thesis is entitled “Optimization and Empirical Analysis of an Event Detection Software for Water Quality Monitoring“. He applied SPOT to a water contamination warning system, which is developed by the US Environmental Protection Agency and Sandia National Laboratories. Martin concludes that “…the automated tuning with SPOT is not only most profitable for the detection performance. It also shows that noteworthy differences exist between the various algorithms that were tuned.” The thesis can be downloaded here.
Martin joined the SPOTSeven team in 2009. He is managing the BMBF project Computational Intelligence basierte Mehrzieloptimierungsverfahren (CIMO). This is not Martin’s first award ceremony: In 2010, he was awarded with the “Erzquell Förderpreis” for his outstanding diploma thesis. Continue reading
The Comprehensive R Archive Network
Allstate compares SAS, Hadoop and R for Big-Data Insurance Models. Conclusion: Revolution R Enterprise gets the same results as SAS, but about 50x faster. Read the full article on R-bloggers. Since there are some problems with this link, here is the link to the original article: http://blog.revolutionanalytics.com/2012/10/allstate-big-data-glm.html
The SPO toolbox can generate detailed reports. A simulated annealing algorithm was tuned in this study. The source code for generating these plots is available below.
My presentation “Research Topics in Sequential Parameter Optimization” is available on-line. It presents noise-handling methods such as Optimal Computing Budget Allocation (OCBA) that are implemented in the Sequential Parameter Optimization (SPO) framework. SPO implements a sequential approach for constructing a meta-model based on noisy data.
ESF members discussing during the coffee break
The presentation was given during the ESF Workshop “Uncertainty Management in Simulation-Optimization of Complex Systems: Algorithms And Applications” in Rome, May 2012.
ESF team: 16 members from 9 countries
You can find the R code, which was used to generate this presentation, below:
…the Big Beacon Manifesto from David Goldberg.
Technology, Entertainment, Design.
It is relatively easy getting a scientific paper published, which reports a positive result, e.g., “My new algorithm A performs significantly better than state-of-the-art optimization algorithm B”.
But, have you ever tried to publish a paper which reports a failure, e.g., “My new algorithm A performs significantly worse than state-of-the-art optimization algorithm B”?
Scientific papers with negative results are rare. Ben Goldacre explains in this TED talk why ignoring negative results can be dangerous—at least in the medical world. Although consequences in math and computer science might not be as life-threatening as in medicine, these disciplines might benefit from reporting negative results as well.
Prof. J.P.C. Kleijnen
The First Workshop On Applied Meta-Modeling is a good opportunity to meet experts in the field of simulation and optimization. In this post, we will present some information about Prof. Jack P.C. Kleijnen.
J. P.C. Kleijnen is Professor of ‘Simulation and Information Systems’ at Tilburg University, where he is a member of the ‘Department of Information Management’ and the ‘Operations Research’ group of CentER of the ‘ Tilburg School of Economics and Management (TiSEM)’.
Warning: the following comic might contain strong language (which may be unsuitable for children), unusual humor (which may be unsuitable for adults), and advanced mathematics (which may be unsuitable for liberal-arts majors).
Members of the PhD committee: Mark Hoogendoorn, Dirk Thierens, Holger Hoos, Thomas Bartz-Beielstein, Gusz Eiben
Prof. Bartz-Beielstein was invited to serve as a member of Selmar Smit’s dissertation committee. Selmar’s thesis “Parameter Tuning and Scientific Testing” was supervised by Prof. Gusz Eiben (Vrije Universiteit Amsterdam).
Announcement in the entrance hall
The thesis is about parameter tuning. Selmar addresses the issue of evaluating and comparing evolutionary algorithms. He promotes a scientific testing approach, in which the focus is on why and when an algorithm performs better, rather than if it is better. He claims that such an approach is not only more informative for the users of algorithms, but also changes the work of researchers from development into science.
First Workshop On
Cologne University of Applied Sciences, Gummersbach
Friday, 16 November 2012, 11:00 – 16:00
Continuing the tradition started at Tilburg University, Netherlands, in 2005 we are happy to host the First Workshop on Applied Meta-Modeling in Gummersbach!
Prof. J.P.C. Kleijnen
Prof. Tom Dhaene
Prof. Dick den Hertog
The schedule looks as follows:
- 11.00 -11.10 Welcome
- 11.10 -11.50 “Beyond the Particular Problem Instance: Generating Problem Classes and Analyzing Algorithm Performance” by Prof. Thomas Bartz-Beielstein, Cologne University of Applied Sciences, Germany
- 11.55 -12.35 “Simulation optimization via bootstrapped Kriging: survey” by Prof.Jack P.C. Kleijnen, Department of Information Management / CentER, Tilburg University, the Netherlands
- 12.35 -13.30 – Break
- 13.30 -14.10 “Surrogate Modeling of Computationally Expensive Black-Box Systems” by Dr. Dirk Deschrijver and Prof. Tom Dhaene, Department of Information Technology, Ghent University, Belgium
- 14.15 -15.25 “Adjustable Robust Parameter Design using Metamodels” by Ihsan Yanikoglu, Prof. Dick den Hertog, Prof. Jack P.C. Kleijnen, CentER, Tilburg University, the Netherlands
- 15.25 -16.00 Discussion
Dr. Dirk Deschrijver
Dr. Katya Vladislavleva
This workshop is organized by Dr. Katya Vladislavleva and the SPOTSeven Team (www.spotseven.org).
Participation is free, but registration is required. To register, please send an email to workshop2012 at gm.fh-koeln.de with a subject “Workshop on Applied Meta-Modeling”.
Springer’s exhibition booth
Promoting “Experimental Methods for the Analysis of Optimization Algorithms”
On Saturday, Prof. Bartz-Beielstein visited the Frankfurt Book Fair. The stay at Springer’s exhibition stand was a great pleasure: it was spacious and very well designed.
The book “Experimental Methods for the Analysis of Optimization Algorithms” could be accessed on computer terminals. Many thanks to Ronan Nugent from Springer for this invitation.