Forwarded mail.... (fwd)

Mie Nov 3 11:19:17 ART 1999

Se invita a los investigadores, docentes y alumnos de distintas
disciplinas que esten interesados en el tema.

Para mayor informacion consultar en Departamento de Informatica o en
Proyecto 338403.

R. Gallard

=========================================================================

    Red Argentina de Cursos de Posgrados en Ciencias de la Computación 

				FOMEC 1999

		 COMPUTACION EVOLUTIVA EN LA ARGENTINA

El Departamento de Informatica de la Universidad Nacional de San Luis
informa que durante el mes de Noviembre, se reuniran en San Luis tres de
los mas destacados investigadores del nuevo campo de la Computacion
Evolutiva. Ellos son:

* El Profesor Zbigniew Michalewicz, de la University of North Carolina at
Charlotte (USA), quien visitara nuestra universidad en calidad de
consultor, como director de cuatro tesis doctorales de nuestro
Departamento. Su permanencia se estima entre el 15 y 26 de Noviembre
proximo.

* El Profesor Marc Schoenauer del Centre de Mathematiques Appliquees Ecole
Polytechnique, 91128 Palaiseau Cedex - FRANCE, quien entre el 8 y 19 de
Noviembre proximo dictara el curso "Evolutionary Optimization -
Applications to Numerical Problems".

* El Profesor Thomas Back del Centrum fur angewandte Systemanalyse (CASA)
Informatik Centrum Dortmund Joseph-von-Fraunhofer-Str. 20 D-44227 Dortmund
and Department of Computer Science Leiden University Niels Bohrweg 1
NL-2333 CA Leiden, quien entre el 22 y 27 de Noviembre proximo dictara el
curso "Lectures on Evolution Strategies: Algorithms, Theory and
Applications".

Tambien se ha invitado a asistir a las reuniones que oportunamente se
programen a nuestra colega, la Profesora María Cristina Riff Rojas de la
Universidad Técnica Federico Santa María, Chile, quien durante nuestro
ultimo CACIC en Tandil, estuvo a cargo del curso "Optimización
combinatoria usando algoritmos evolucionistas". Se espera contar con su
visita en esos dias.

Informacion sobre otras actividades a desarrollarse (seminarios y/o
conferencias) se dara a publicidad oportunamente.

Este mensaje esta dirigido a investigadores, docentes o alumnos de
posgrado que esten trabajando en estos temas o deseen hacerlo
estableciendo contacto con los investigadores mas destacados del area.

A continuacion se detallan los cursos y antecedentes de los responsables
de los mismos.

=============================================================================

"Evolutionary Optimization - Applications to Numerical Problems".  
Dr.Marc Schoenauer

(8/11/99 al 19/11/99). Cuatro a seis horas diarias. Aprobacion por
evaluacion escrita o por medio del desarrollo de un proyecto breve (2 a 4
semanas luego de finalizado el curso).

Topics: 
* Bases of EC, historical roots. 

* Theoretical results: From schema theory to Markov chain analysis for
binary representations, some convergence results for real-valued
representations.

* The modern trends: representation and operators, forma theory, from
binary to real representation.

* Advanced EC: Multimodal functions, constrained optimization,
multi-objective problems, parallel algorithms, hybridization of EC and
local search.

* All points will be illustrated by numerous applications, including
combinatorial optimization (from TSP to mesh re-numbering), parameter
optimization (electromagnetism, optics), mixed problem (optical filter
design). Special emphasis will be put on non-parametric optimization, for
inverse problem solving that amount to function identification (structural
mechanics, chemical engineering, ...)

Dr. Marc Schoenauer CV: 

Studies: Graduated at Ecole Normale Superieure in Paris. PhD (French
regulation) in applied maths in 1980.

Position: Full time researcher at CNRS since 1980 ("charge de recherche
1ere classe). Head of the EEAAX - Artificial Evolution and Machine
Learning group at Ecole Polytechnique.

Teaching: Many courses on numerical analysis. Course on Evolutionary
Computation in the post-graduate studies in applied maths (Ecole
Polytechnique and University Paris 6), and in the Post-graduate studies in
Mathematical Engineering (Ecole Polytechnique - EPFL).

Conferences organization and other scientific activities: Founding
president of the French Society for Artificial Evolution, which organizes
the bi-annual series of conference Artificial Evolution (Springer Verlag
LNCS 1063 and 1363). Member of numerous Program Committee (PPSN, ICGA and
IEEE series), was program chair of last PPSN (Amsterdam 98) and wil be the
general chair of the joint IEEE-ICEC and PPSN in year 2000 in Paris.
Member of the editorial board of IEEE Transactions on Evolutionary
Computation, and of the IEEE Technical Committee on Evolutionary
Computation.

==================================================================
Lectures on Evolution Strategies: Algorithms, Theory and Applications. 

Dr. Thomas Back. 

(22/11/99 al 27/11/99). Cuatro a seis horas diarias. Aprobacion por
evaluacion escrita o por medio del desarrollo de un proyecto breve (2 a 4
semanas luego de finalizado el curso).

Evolution strategies are a class of search and optimization algorithms
gleaned from the model of natural evolution. Together with e.g. genetic
algorithms, evolutionary programming, and genetic programming, evolution
strategies are instances of the class of evolutionary algorithms. While
all of these algorithms share the common metaphor of a population of
candidate solutions (to a given search or optimization problem) evolving
by iterated processes of selection and variation by means of mutation and
recombination, their realizations are remarkably different. This lecture
series focuses on evolution strategies, an instance of evolutionary
algorithms invented in the mid 60s and continuously extended over the past
thirty years. The most distinguishing features of modern evolution
strategies are the utilization of a real-valued representation of
candidate solutions, normally distributed mutations with self-adaptive
variances (and covariances), a deterministic environmental selection
operator discarding most of the offspring, and recombination operators
based on the exchange or averaging of object variables. In particular, the
self-adaptation of strategy parameters facilitates an on-line evolution of
strategy parameters in a way similar to the evolution on the level of
candidate solutions. The most important variants of the evolution
strategy, ranging from the (1+1)-strategy to the (&mu+&lambda)- and
(&mu,&lambda)-strategy, are presented in the lectures from a historic,
algorithmic and theoretical point of view. The variation and selection
operators of these variants are discussed, with a particular emphasis on
the self-adaptation of strategy parameters. The theoretical properties of
evolution strategies are also briefly presented to summarize the main
results of the analysis. Moreover, the lectures also present experimental
results of evolution strategies on a number of test problems and give some
examples of practical applications to industrial problems. Other topics of
the lectures include a general introduction to evolutionary computation,
the definition of the terminology used in optimization, and the
explanation of the relationship between genetic algorithms, evolutionary
programming, and evolution strategies.

The contents of the lectures is briefly outlined in the following
overview:  

1.Introduction to evolutionary computation: Biological
background, general outline of an evolutionary algorithm, advantages of
evolutionary computation. 

2.Introduction to optimization: Optimization
problem, local and global optima, constraints.  

3.Evolution strategies I: History, (1+1)-strategy, (_+1)-strategy,
1/5-success rule, evolution window, first applications.

4.Evolution strategies II: (_+*)-stragey,
(_,*)-strategy, mutation operator, recombination operators, selection
operators, outline of selfadaptation.  

5.Evolution strategies III: Self-adaptation, one and n variances,
covariances, experimental and theoretical investigations.

6.Evolution strategies IV: Theoretical analysis, convergence velocity,
global convergence with probability one (main results only).  

7.Evolution strategies V: Extensions and modifications of the basic
algorithm, multi-population strategies, modified self-adaptation, parallel
strategies.

8.Evolution strategies VI: Practical application examples, academic and
industrial test problems, practical demonstration.

9.Evolution strategies VII: Noisy objective functions, multiple criteria
decision making, evolution with subjective selection.  

10.Comparison of evolution strategies, genetic algorithms, and
evolutionary programming. As supporting material for the lecture, the
transparencies will be made available to the audience. For a deeper
understanding of the material presented, we will recommend further reading
in due time.

Dr. Thomas Back CV: 

Thomas Back received the Diploma degree in Computer Science in 1990 and
the Ph.D. degree in Computer Science in 1994, both from the University of
Dortmund, Germany. In 1995, he received the best dissertation award of the
Ger-man Association for Computer Science (GI) for his Ph.D. thesis on
Evolutionary Algorithms. From 1990-1994, he worked as a Scientific
Assistant at the Department of Computer Science of the University of
Dortmund. Since 1994, he is Senior Research Fellow at the Center for
Applied Systems Analysis within the Infor-matik Centrum Dortmund, and
Managing Director of the Center for Applied Systems Analysis since 1996.
He also serves as an Associate Professor in the Computer Science
Department of Leiden University, The Netherlands, and teaches courses on
evolutionary computation at the University of Dortmund and at Leiden
University. His current research interests are in the areas of theory and
application of evolutionary computation and related areas of
computa-tional intelligence. He is author of the book Evolutionary
Algorithms in Theory and Practice: Evolution Strategies, Evolutionary
Programming, Genetic Algorithms (New York, NY: Oxford University Press,
1996), co-editor-in-chief of the Handbook of Evolutionary Computation (New
York, NY: Oxford University Press & Institute of Physics Publish-ing,
Bristol, 1997), associate editor Evolutionary Computation (Cambridge, MA:
The MIT Press), and associate editor of the IEEE Transactions on
Evolutionary Computation. Dr. Back is a member of the IEEE and the Dutch
Association for Theoretical Computer Science (NVTI), serves on the IEEE
Neural Networks Council's technical committee on evolutionary computation
since 1995, was a co-program chair of the 1996 and 1997 IEEE International
Conferences on Evolutionary Computation (ICEC) and the Fifth An-nual
Conference on Evolutionary Programming (EP '96), and was program chair of
the Seventh International Confer-ence on Genetic Algorithms and Their
Applications (ICGA '97).

Su posicion actual es la siguiente: 

Managing Director Center for Applied Systems Analysis (CASA) Informatik
Centrum Dortmund (ICD) Joseph-von-Fraunhofer-Str. 20, D-44227 Dortmund,
Germany Associate Professor Department of Computer Science, Leiden
University Niels Bohrweg 1, NL-2333 CA Leiden, The Netherlands
===============================================================