Genetic Algorithms Digest Wednesday, December 13, 1995 Volume 9 : Issue 61 - Do NOT send email or reply to gadistr@aic.nrl.navy.mil (GA List Moderator) - Send submissions (articles) to GA-List@AIC.NRL.NAVY.MIL - Send administrative (subscribe, unsubscribe, change of address, etc.,) requests to GA-List-Request@AIC.NRL.NAVY.MIL ****************************************************************************** - You can access back issues, GA code, conference announcements, etc., either through the WWW at URL http://www.aic.nrl.navy.mil/galist/ or through anonymous ftp at ftp.aic.nrl.navy.mil [192.26.18.68] in /pub/galist. ****************************************************************************** Today's Topics: - GA Post-Graduate Programs (Re: v9n59) - A Question - Decreasing mutation rate (Re: v9n60) - New thesis available at TCGA - Adpative systems preprint by FTP - MENDEL'96 - CFP - Questions ---------------------------------------------------------------------- CALENDAR OF GA-RELATED ACTIVITIES: (with GA-List issue reference) SAC96 Symposium on Applied Computing GA & Opt Track, PA (v9n33) Feb 17-19, 96 Appl of Fuzzy Technologies in Business and Industry, Germany (v9n54)Feb 29,96 EP96 5th Conf on Evol Programming, San Diego, CA (v9n18) Feb 29-Mar 3, 96 CIFEr96 Comp Intelligence for Financial Eng, NY City (v9n53) Mar 24-26, 96 AAAI-96 Spring Symposium Series, Stanford University, CA (v9n44)Mar 25-27, 96 ACEDC96 Adaptive Computing in Eng. Design & Control, UK (v9n28) Mar 26-28, 96 SOCO96 Intl Symposia w/ Workshops/Tutorials, Reading, UK (v9n35)Mar 26-28, 96 AISB96 Workshop on Evolutionary Computing, Sussex, UK (v9n54) Apr 1-2, 96 SECTAM96 AI Techniques in Eng and Mechanics, Tuscaloosa (v9n46) Apr 14-16, 96 ALifeV Artificial Life Conference, Nara, Japan (v9n45) May 16-18, 96 ICEC96 IEEE Intl Conf on Evol Comp, Nagoya, Japan (v9n18) May 20-22, 96 ISRAM96 Session on Evol Algs in Robotics, Montpellier (v9n46) May 27-30, 96 EvCA96 Evol Comp and Its Applications, Moscow, Russia (v9n59) Jun 24-27, 96 MENDEL96 2nd Intl Mendel Conference on GAs, Brno, Czech (v9n61) Jun 26-28, 96 IPMU96 Granada, Spain (v9n31) Jul 1-5, 96 ICML96 Intl Conf on Machine Learning, Bari, Italy (v9n54) Jul 3-6, 96 GP96 Genetic Programming Conference, Stanford, CA (v9n9) Jul 28-31, 96 FOGA4 Foundations of Genetic Algorithms, San Diego, CA (v9n47) Aug 3-5, 96 2NWGA 2nd Nordic Wrkshp on Genetic Algorithms, Finland (v9n56) Aug 19-25, 96 EUFIT96 Intelligent Techniques and Soft Computing, Germany (v9n60)Sep 2-5, 96 SAB96 From Animals to Animats, Cape Cod, Massachusetts (v9n31) Sep 9-13, 96 PPSN96 Parallel Problem Solving from Nature, Berlin (v9n17) Sep 22-27, 96 ICGI96 Colloquium on Grammatical Inference, Montpellier (v9n45) Sep 25-27, 96 (Send announcements of other activities to GA-List@aic.nrl.navy.mil) ------------------------------ From: William M. Spears (Moderator) Date: Tuesday, Dec 12 Subject: GA Post-Graduate Programs (Re: v9n59) I received the following replies about GA post graduate programs: >From: gunho@hp832.informatik.hu-berlin.de > > The dept. of Evolution Strategy at the Technical University of Berlin offers > a program in GA/ES. Right now 6-8 PhD stuents work on different subjects. > See also their homepage: > > http://lautaro.fb10.tu-berlin.de/ > > cheers, > Guenther >From: Robert Elliott Smith > > I have a grad program in GAs. > > Take Care > > Robert Elliott Smith > Department of Engineering Science and Mechanics > Room 210 Hardaway Hall > The University of Alabama > Box 870278 > Tuscaloosa, Alabama 35487 > <> rob@comec4.mh.ua.edu > <> (205) 348-1618 > <> (205) 348-7240 > <> > http://hamton.eng.ua.edu/college/home/mh/faculty/rsmith/Web/smith.html >From: Peter Ross > >In v9n59 you asked about post-grad programs in GAs. We have a thriving >group of MSc and PhD students studying a good range of topics. >See http://www.dai.ed.ac.uk/ for more information about the >department, and click your way to our evolutionary computing group >page. >From: dasgupta@arch.umsl.edu (Dipankar Dasgupta) > > Here is some information regarding Graduate Studies in Genetic Algorithms. >The URL location is > http://chinook.uoregon.edu/~ben/ga-grad.html > >Hope that will help. >Thanks, >-Dipankar I have added those links to my home page: http://www.aic.nrl.navy.mil/~spears If anyone has any more pointers of that nature that they wish me to add, please send email to me at spears@aic.nrl.navy.mil. Thanks, Bill ------------------------------ From: "Harold Gibbs" Date: Thu, 07 Dec 95 12:57:10 PST Subject: A Question As a relative newcomer to GA's, I have a question that will hopefully be easily answered by someone more experienced. I am trying to use GA's to generate rules to classify noisy time series but am experiencing problems with overfitting the data. What are the generally accepted approaches to this problem in the GA field? I know that in the Neural Network arena this topic is widely discussed and researched. Can anyone refer me to any relevant articles or research papers? Harold Gibbs Harold_Gibbs@county.com.au ------------------------------ From: efalkena@ulb.ac.be (Falkenauer Emanuel) Date: Thu, 7 Dec 95 03:06:20 +0100 Subject: Decreasing mutation rate (Re: v9n60) Hello, I have seen Bill Hart's call for references where the mutation rate is varied during GA search, and in particular those where the rate is 'annealed' towards zero (GA-List v9n60, "References for GAs with variable mutation probabili- ties"). Among the references Bill has already gathered is Terrence Fogarty's ICGA'89 paper "Varying the probability of mutation in the genetic algorithm". That is an extremely good paper, but I would like to point out at least one occasion where it has been thoroughly misused. Some time ago, I came across a paper published by the Operations Research Society of America, where the authors used a mutation scheme that effectively reduced the mutation rate to nil over the generations. The reference 'justify- ing' that approach was the ICGA'89 paper. I am open to any idea, but somehow I found the approach fascinating at that time, and so I got hold of the paper and actually read it. It turned out that Terry artificially increased the mutation rate in the begin- ning of the GA search for the simple reason that was starting with a completely *uniform population* (all individuals identical, corresponding to a good solution to the optimization problem being solved). Terry's nice idea was clear: since he already had a good solution, he 'cloned' it over the whole population. However, he faced the obvious problem of introdu- cing diversity, so that the GA could take off. And since there are many ways of 'perturbing' a solution (many of which are bad), he let the care of choosing the right one(s) to the survival mechanism, and that scheme yielded nice re- sults. Perhaps unsurprisingly, Terry also noted in that paper that when he started with a randomly generated population (as is usually the case in a GA), he did NOT observe an advantage in lowering the mutation rate. The sad point is that the paper that referenced Terry of course generated the initial population at random. Still, they claimed that since lowering of the mutation rate was good for Terry, then it was a good strategy to apply in their setup as well. To be honest, I wondered how that paper ever made it through the referees. I'm sure that there are probably good reasons for varying the mutation rate and, although I'm personally inclined to increase it over the generations, there are maybe some reasons for doing exactly the opposite (i.e. lower it). However, as with any idea, please let's be careful not to take it outside the context in which it applies. Cheers, Emanuel. ------------------------------ From: Robert Elliott Smith Date: Tue, 05 Dec 95 17:30:48 -0600 Subject: New thesis available at TCGA The Clearinghouse for Genetic Algorithms at The University of Alabama is pleased to announce the availability of the following Master's thesis. The thesis can be obtained by requesting a hardcopy from the address at the bottom of this message. Note that TCGA requires a check or money order for $9.00 US ($12.00 US for overseas mail) to defray the cost of shipping and handling of theses or dissertations. You can also obtain this TCGA report (and some others) through the TCGA homepage. This page is still a work-in-progress, so be patient! http://galab3.mh.ua.edu:1025/tcga.html New Thesis: Cooperative Learning Classifier Systems: A Neural Network Approach", TCGA Report No. 95002 University of Alabama Tuscaloosa (Master's Thesis) ABSTRACT The learning classifier system (LCS) is a rule-based system capable of adapting to its environment. Architectural descriptions of the LCS can take on the guise of a network structure. An preliminary investigation into the links between the LCS and ANN is made. The nature of an artificial neural LCS is explored. This study presents the initial results for a batch update and an incremental update neural network LCS. The two systems are described in depth, as to provide for implementation. The two systems are tested on the six multiplexor [Wilson, 1986, 1990]. Results illustrate that the batch version converges to concise rule-set with desirable performance. The second system, the incremental version, lends itself readily to large state spaces with single input-single update constraints. A discussion of future directions concludes the study. References [Wilson:86] Wilson, S.W. (1986) Classifier Systems and the Animat Problem. (Research Memo RIS No. 36r) Cambridge, MA: The Rowland Institute of Science. [Wilson:90] Wilson, S.W. (1990) Perceptron Redux: Emergence of structure. In S. Forrest (Ed.) Emergent Computation: Proceedings of the Ninth Annual International Conference for Nonlinear Studies on Self-Organization, Collective, and Cooperative Phenomena in Natural and Artificial Computing Networks, 249--256, Amsterdam. North-Holland. Contact TCGA through: Robert Elliott Smith Department of Engineering Science and Mechanics Room 210 Hardaway Hall The University of Alabama Box 870278 Tuscaloosa, Alabama 35487 <> rob@comec4.mh.ua.edu <> (205) 348-1618 <> (205) 348-7240 <> http://hamton.eng.ua.edu/college/home/mh/faculty/rsmith/Web/smith.html ------------------------------ From: Steve Frank Date: Mon, 11 Dec 95 14:34:47 -0800 Subject: Adpative systems preprint by FTP Adpative systems preprint by FTP The following preprint is available for ftp: Frank, S. A. 1996. The design of natural and artificial adaptive systems. In Evolutionary Biology of Adaptation, M. R. Rose and G. V. Lauder, eds. Academic Press (in press). The paper covers some familiar topics, but veterans of the field may find useful my emphasis of new research problems in biology, computing and engineering. I have also attempted to list and explain concisely the essential features of adaptive systems. I found this surprisingly difficult at first, and would like to hear criticisms of my current outline and related examples from biology and computing. The first few pages of the paper can be viewed at http://jojo.bio.uci.edu/usr/frank/homepage/abstracts/96Adapt.html The paper is available in a variety of formats. The best place to start is on the web page http://jojo.bio.uci.edu/usr/frank/homepage/reprints.html#96Adapt under the item 96Adapt. Those with ftp access only will probably want to download ftp://jojo.bio.uci.edu/pub/steve/reprints/1996/96Adt-nf.ps.gz which is 264KB and is missing three figures or ftp://jojo.bio.uci.edu/pub/steve/reprints/1996/96Adapt.ps.gz which has all figures but is much larger at 3800KB. Steven Frank | Dept. of Ecology and | Tel: 714-824-2244 Evolutionary Biology | Fax: 714-824-2181 Univ. of California | email: safrank@uci.edu Irvine, CA 92717 | ------------------------------ From: "Ivan Simonik,st. 91/96" Date: Mon, 11 Dec 1995 15:13:02 CET Subject: MENDEL'96 - CFP [ WMS: The full article was too long for GA-List. Please see pub/galist/info/ conferences/MENDEL96 on ftp.aic.nrl.navy.mil for more info. Also see: http://www.fme.vutbr.cz/html/UAI/confs/men96.html ] MENDEL '96 2nd International Mendel Conference on Genetic Algorithms June 26 - 28, 1996 BRNO, CZECH REPUBLIC In 1865 Gregor Johann Mendel, an Augustinian priest in a Brno monastery, described to the Brno Natural Science Society the transfer of determinants in pea plants. Sadly, the fundamental importance of Mendel's finding was not understood by the society. Until 1900, nobody was realized that Mendel had discovered the Law of Heredity. Brno, the workplace of the father of genetics, and the conference organisers, are very pleased to offer a conference platform for modern engineering using heredity principles. The MENDEL '96 Conference organisers wish to create a base for a theoretical and practical exchange of knowledge of evolution theories in Central and Eastern Europe. Theoretical papers that relate to the conference topic are invited as are any practical results and knowledge from the field of evolution strategies in mathematics, engineering, research, education, etc. SCOPE The scope of the Conference includes, but is not limited to Innovation and Theory - genetic operators and representations - hybrid and distributed algorithms - genetic algorithms for neural networks - parallel genetic algorithms - genetic programming - evolutionary programming - mathematical techniques and analysis - biological aspects of GAs Applications - optimisation problems - machine learning - scheduling algorithms - signal processing - robotics - vision system/applications - process industry - control systems - manufacturing - system identification - decision-making - economics - biomedical The Conference target workers in the field of Engineering Systems or related fields in Mathematics, Physics, Computer Science, etc. Workers in other areas such as Biology, Psychology and Medicine are also welcomed. WORKING LANGUAGE The working language of the Conference will be English which will be used for all printed materials, presentations and discussions. DEADLINES February 28, 1996 Submission of paper (original and 3 copies) April 15, 1996 Notification of acceptance to authors April 30, 1996 Final Announcement distribution Submission of papers should be addressed to: MENDEL '96 Secretary: Technical University of Brno Faculty of Mechanical Engineering Institute of Automation and Information Technology Technicka 2 616 69 Brno Czech Republic Phone: +42-5-41143334 Fax: +42-5-744979 e-mail: mendel96@kinf.fme.vutbr.cz Other information will be available on WWW address "http://www.fme.vutbr.cz/html/UAI/confs/men96.html" ----------------------------- From: pharizna@repsol.es Date: Wed, 13 Dec 95 16:38:03 GMT Subject: QUESTIONS 1. How can I get some information about GA applications in Chemical Engineering (simulation and control in Petrochemical Process)? 2. Are Genetic Algorithms been using to build Inferential Models in order to have correlation between Process Conditions vs Chemical composition ? 3. Are there any GA package in order to "build" Neural Networks? ------------------------------ End of Genetic Algorithms Digest ******************************