2012-02-13: Distributed Computing

Distributed Computing Distributed Systems Teaching

This Tuesday (2012-02-14), I will start teaching the lecture Distributed Computing at the School of Computer Science and Technology here in Héféi. The course home page is http://www.it-weise.de/teaching/2012SS-DC.

The course will not follow a book, so as teaching material, we will mainly use the course slides. This gives us freedom to tackle a set of different interesting topics in distributed computing with hands-on-experience. I will, e.g., try to do some lessons on MPI, some on cloud computing and Map/Reduce, discuss SOA and web services, as well as fundamental stuff such as sockets. I hope that I can make this course interesting and useful for all participants and I will put lots of work into that.

2011-11-18: Book Now Online

Applications Evolutionary Algorithms Publications

Finally, the book "Variants of Evolutionary Algorithms for Real-World Applications" [d4] which Raymond Chiong and Zbigniew Michalewicz and I edited together (blog:2011-09-30) has appeared online in springerlink. Its doi is 10.1007/978-3-642-23424-8 and it is available under the ISBN numbers 978-3-642-23423-1 and 978-3-642-23424-8. It's Google Books ID is B2ONePP40MEC and you can order it from Amazon under ID 3642234232.

2011-10-26: CEC Special Session: Large Scale Global Optimization

Evolutionary Algorithms Heuristics Large-Scale Publications

I am very happy to announce the "CEC 2012 Special Session and Competition on Large Scale Global Optimization" which I have the honor to co-chair with my valued colleagues Ke Tang and Zhenyu Yang.

The 2012 IEEE Congress on Evolutionary Computation (IEEE CEC 2012) will take place as part of the 2012 IEEE World Congress on Computational Intelligence (IEEE WCCI 2012) from June 10 to 15, 2012 in the International Conference Center of Brisbane in Australia.

Call for Participation

In the past two decades, different kinds of nature-inspired optimization algorithms have been developed and applied to solve optimization problems, including Simulated Annealing (SA), Evolutionary Algorithms (EAs), Differential Evolution (DE), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Estimation of Distribution Algorithms (EDA), etc. Although these approaches have shown excellent search abilities when applying to some small or medium size problems, many of them will encounter severe difficulties when applying to large scale problems, e.g., problems with up to 1000 variables. The reasons appear to be two-fold. First, the complexity of a problem usually increases with the number of decision variables, number of constraints, or even number of objectives (for multi-objective optimization). This emergent complexity might prevent a previously successful search strategy from finding the optimal solution. Second, the solution space of the problem increases exponentially with the number of decision variables, and a more efficient search strategy is required to explore all the promising regions with limited computational resources.

Historically, scaling up EAs to large scale problems has attracted much interest, including both theoretical and practical studies. However, existing work in the areas of EAs are still limited given the significance of the scalability issue. Due to this fact, this special session is devoted to highlight the recent advances in EAs for large scale optimization problems, involving single objective or multiple objectives, unconstrained or constrained problems, binary or discrete or real or mixed decision variables. Specifically, we encourage interested researchers to submit their latest work on:

Both theoretical and experimental analysis of the scalability of EAs.
Novel approaches and algorithms for scaling up EAs to large scale optimization problems.
Applications of EAs to real-world large scale optimization problems.
Papers on novel test suites that help us in understanding problem characteristics are also welcome.

Furthermore, a competition on Large-Scale Numerical Optimization will also be organized in company with our special session. This competition is built on the successful special session and competition on LSGO in CEC?2010. For the competition in CEC?2012, the previously proposed CEC?2010 benchmark test suite, which consists of 20 benchmark test functions capturing a range of problem characteristics, will be used. Participants of the competition will be required to evaluate their existing or novel algorithms using the test suite, and are welcome to report their approaches and results in a paper submitted to the special session. The competition will provide the participants a great opportunity to compare their LSGO algorithms with others.

For more information about the special session and competition, such as the deadlines and the benchmark problems, please visit the corresponding webpages:

For Special Session: http://staff.ustc.edu.cn/~ketang/cec2012/cec2012lsgo.htm
For Competition: http://staff.ustc.edu.cn/~ketang/cec2012/lsgo_competition.htm

Special Session and Competition Organizers

Ke Tang
Nature Inspired Computation and Applications Laboratory (NICAL),
School of Computer Science and Technology,
University of Science and Technology of China, China
ketang@ustc.edu.cn
Zhenyu Yang
Department of Computer Science and Technology
East China Normal University, Shanghai, China
zhyyang@cs.ecnu.edu.cn
Thomas Weise
Nature Inspired Computation and Applications Laboratory (NICAL),
School of Computer Science and Technology,
University of Science and Technology of China, China
tweise@ustc.edu.cn, http://www.it-weise.de/

2011-10-15: Multiobjectivization via Helper-Objectives with the Tunable Objectives Problem

Evolutionary Algorithms Multi-Objective Publications

Today, the paper Multiobjectivization via Helper-Objectives with the Tunable Objectives Problem [1] became visible as online preview on the journal website. This paper investigates the utility multi-objectivization (solving single-objective problems as multi-objective problems [2, 3, 4]) by using an abstract benchmark problem. This benchmark problem — Tunable Objectives Problem (TOP) — allows to build problems with different degrees of objective-convolution, multiple layer epistasis, and the presence of local optima.

The TOP is based on one of the works [d56] which my colleagues and I did back in my time as PhD student at the Distributed Systems Group at the University of Kassel in Germany. You can find the corresponding paper, presentation slides, and the Java implementation contributed by my former Student Stefan Niemczyk under [d56] in my publications list. Stefan wrote his Diplom I Thesis (in German language) about this topic, you can find it under [d59] in my documents list. It is really nice to see that our work did not disappear in the Nirvana but actually became useful for some colleagues. Even more, Lochtefeld and Ciarallo did not just use our work, they improved it significantly. The TOP is a really nice and worthy successor of our tunable problem, even with more capabilities.

The search space of both the TOP and our tunable problem are bit strings of length n. The problems are therefore ideal to test, e.g., Genetic Algorithm. They are composed of layered transformations. In each layer, a candidate string is transformed according to some rules and finally compared to an ideal solution. Its objective value will be the Hamming distance to that solution. Both problems are designed to be used for multi-objective optimization. The idea is that each transformation layer can introduce one specific facet of problem hardness: one transformation may simulate neutrality/redundancy, another one may introduce epistasis, a next one may add ruggedness or deceptivity, then objective convolution may be introduced, another layer may provide a simulation for noise or affinity for overfitting. Each layer can be tuned, i.e., we can tune its corresponding problem difficulty facet either to a maximum, turn it off, or choose any value in between. This allows testing an optimization algorithm for many different combinations of problematic difficulties and can give performance information on a detail level which should allow for a clear analysis of algorithm strengths and weaknesses. [1, d56, d59]

[1]	Darrell F. Lochtefeld and Frank W. Ciarallo. "Multiobjectivization via Helper-Objectives with the Tunable Objectives Problem," in IEEE Transactions on Evolutionary Computation (TEVC/IEEE TEC), volume 16, 2012, published by IEEE CIS, doi:10.1109/TEVC.2011.2136345 (available as online preview)
[2]	Joshua D. Knowles, Richard A. Watson, David Wolfe Corne. "Reducing Local Optima in Single-Objective Problems by Multi-objectivization." In Proceedings of the First International Conference on Evolutionary Multi-Criterion Optimization, March 7–9, 2001, Eidgenössische Technische Hochschule (ETH) Zürich, Zürich, Switzerland, pages 269–283, Volume 1993 of Lecture Notes in Computer Science (LNCS), Springer, Berlin. doi:10.1007/3-540-44719-9_19, url:http://www.macs.hw.ac.uk/~dwcorne/rlo.pdf
[3]	Julia Handl, Simon C. Lovell, and Joshua D. Knowles. "Multiobjectivization by Decomposition of Scalar Cost Functions." In Proceedings of 10th International Conference on Parallel Problem Solving from Nature, September 13–17, 2008, Dortmund, North Rhine-Westphalia, Germany, pages 31–40, Volume 5199 of Lecture Notes in Computer Science (LNCS), Springer, Berlin. doi:10.1007/978-3-540-87700-4_4, url:http://dbkgroup.org/handl/ppsn_decomp.pdf
[4]	Martin Jähne, Xiaodong Li, and Jürgen Branke. "Evolutionary Algorithms and Multi-Objectivization for the Travelling Salesman Problem." In Proceedings of the 11th Annual Conference on Genetic and Evolutionary Computation, July 8–12, 2009, Montréal, QC, Canada, pages 595–602, ACM, New York, NY, USA. doi:10.1145/1569901.1569984, url:http://goanna.cs.rmit.edu.au/~xiaodong/publications/multi-objectivization-jahne-gecco09.pdf

2011-10-02: Task Force "Education"

Evolutionary Algorithms Teaching

Today is the launch of the new task force "Education" of the IEEE Computational Intelligence Society, Emergent Technology Technical Committee (IEEE CIS ETTC). I have the great honor to chair this task force, together with my vice chairs Raymond Chiong and Mateen Rizki. The website of the task force is www.emergentCItechs.org.

The application of evolutionary optimization techniques in practice and the industry still meets with skepticism and prejudices. Especially emerging CI technologies face this problem, even in the academic sector. The reasons are that only few people are trained and educated in CI and its emergent fields. It thus is the goal of our task force to provide a wider audience with a clearly organized, sorted, comprehensive, up-to-date, and free collection of education and teaching material on emergent Computational Intelligence-based technologies. We will collect, review, and present a set of high-quality education, teaching, and training material in a central repository. As such material becomes available and easier to find, more students and practitioners will acquire knowledge in the area and it also becomes easier for university teachers to include corresponding topics into their curriculum.

The main goal of this task force is to build a central website and repository (at http://www.emergentCItechs.org) for tutorials, teaching material, how-tos, examples, sources, manuals, and references on Emergent Technologies in Computational Intelligence as listed in section "Scope" below. We will therefore collect and compile videos of tutorials, invited talks, and plenary talks at major CIS conferences, review the online education materials that are related to the interests of the Emergent Technology Technical Committee (ETTC) for their relevance and accuracy, and assist in other CIS educational activities within the fields of ETTC. We will also provide a catalogue on courses, syllabuses, teaching material, literature, software and sources, as well as real-world applications concerning the related fields.

This mission statement sounds really fancy. I hope we can live up it. This is the first time that I am leading a task force or similar activity. I am a bit afraid of whether I will be a good leader. I have to try hard.

...more

2012-02-13: Distributed Computing
2011-11-18: Book Now Online
2011-10-26: CEC Special Session: Large Scale Global Optimization
2011-10-15: Multiobjectivization via Helper-Objectives with the Tunable Objectives Problem
2011-10-02: Task Force "Education"
2011-09-30: Variants of Evolutionary Algorithms for Real-World Applications
2011-09-05: Practical Optimization Algorithm Design
2011-07-26: Travel in China
2011-07-20: Travel Home to Germany
2011-05-30: Wenxiang Chen One of the Winners of the Anhui-level Challenge Cup
2011-05-28: Summer Concert of USTC Orchestra
2011-05-22/23: Visit by my Friend Ingo
2011-05-12: Presentation at the Jiujiang University
2011-04-17: 2011 International Workshop on Nature Inspired Computation and Its Applications
2011-03-27: Good News and Awards
2011-03-20: Research Talk by Prof. Geihs Scheduled on 2011-03-26
2011-03-19: Leslie Vailiant: Turing Award
2011-03-09: Axis2 + JiBX: The New Environment for AOAB
2011-03-07: XML/HTML Entities and LaTeX
2011-03-03: Java + XML + Web Services
2011-03-02: Graphical Visualization of Statistical Test Results
2011-02-26: Web Services with Axis2
2011-02-24: Evolutionary Intelligence Special Issue
2011-02-23: Picbreeder
2011-02-22: LaTeX Packages
2011-02-21: Website Generation

Tags