The idea of this Dagstuhl-Seminar was born during the Third International Conference on Parallel Problem Solving from Nature (PPSN III) at Jerusalem in 1994. Among those methods to solve problems by means of computers using natural metaphors, Evolutionary Algorithms (EAs) are dominant, nowadays. They play an important role in two new fields of computer science called Computational Intelligence (CI) and Artificial Life (AL), which emerged aside of the now classical realm of Artificial Intelligence (AI). Whereas AI uses symbolic knowledge processing techniques, CI and AL emphasize the use of subsymbolic computing, sometimes also called Soft Computing (SOCO).

The terms Evolutionary Algorithms (EAs) and Evolutionary Computing (EC) were coined at the beginning of the 1990ies, i.e., during the first PPSN (Parallel Problem Solving from Nature) conference at Dortmund and the fourth International Conference on Genetic Algorithms (ICGA) at San Diego, in order to have a common denominator for at least three different, independently of each other and at different places but nearly at the same time, the 1960ies, created methods to solve problems, mostly optimization problems, by means of mimicking mechanisms gleaned from organic evolution:

• Evolutionary Programming (EP), first introduced by Lawrence Fogel at San Diego and recently revived by David Fogel;
• Evolution Strategies (ESs), developed by Ingo Rechenberg and Hans– Paul Schwefel at Berlin;
• Genetic Algorithms (GAs), proposed by John Holland at Ann Arbor and first used as an optimization algorithm by Kenneth De Jong.

It took nearly a human generation until these methods became accepted and used for a variety of tasks for which classical methods are insufficient and exact (enumerative) methods too time-consuming. So far, applications of EAs have been more convincing than their theory, which is still weak, though under progress.

There were about 40 presentations which were subsumed under the following topics: theory of EAs, self adaptation, parallelity, application (industry, economics, and others), computational intelligence, genetic programming, and natural and artificial evolution. The participants got deeper insights into methods, mechanisms, foundations, and application areas, but there remained a lot of new questions which we could not answer in this week. Evolutionary Algorithms are forming still a young and rich field of scientific research and application oriented experiments with surprises and disadvantages, where foundations, useful methods, codings, fitness and their landscapes, and experiences by applications and experiments must and will be developed intensively in the next decade.

These considerations led to the idea of emphasizing EA applications during this seminar. Nevertheless, we started with a survey of the state of art concerning theoretical results about the convergence and convergence rates as far as available for EP, ESs, and GAs.

It was a pity that those invited to come here from the U.S.A. finally could not attend due to the budget struggles over there. However, some leading scientists from other European countries came here and provided their recent results and experiences. Though the German scene was not completely present, as well, many groups from academia as well as industry sent delegates and thus contributed to the exchange of knowledge and ideas as well as many lively discussions about open questions.

The ambients of this unique place, i.e. Schloß Dagstuhl, was optimal, indeed, to take a step forward into the exciting field of Evolutionary Algorithms and — may be even more important — towards closer personal connections between those interested in it.

Creative Commons BY 3.0 DE