http://evohybrid04.dei.uc.pt

This workshop will focus on the application of hybrid evolutionary algorithms (EAs) to complex optimization problems. Standard EAs often perform poorly when searching for good solutions for complex problems and may benefit if they are combined with other techniques. Broadly speaking, we can consider two large classes of hybrid architectures:

- The EA can be complemented with a local and/or deterministic search method. The joint application of both techniques provides a trade-off between stochastic global exploration and fine-grained exploitation.

- The EA can be enhanced with problem specific heuristics adding explicit knowledge about the problem being solved. Adopting several current approaches as a starting point, this workshop aims at promoting a widespread discussion about this topic and, most important, to analyze if it is possible to develop new hybrid architectures that perform better than today's methods.

The themes of the workshop include, but are not restricted to:

- Application of hybrid evolutionary approaches to complex optimization problems;
- Common hybridization techniques, such as local improvement of candidate solutions, intelligent chromosome decoders or heuristic variation operators;
- Hybridization of EAs with state-of-the-art techniques frequently used in the optimization of complex problems. Linear programming and branch-and-cut are two examples of such methods;
- Analysis of the strengths (and weaknesses) of today's hybrid approaches. How do they compare to other techniques that are also applied in such problems?
- Promising directions for future research.

Military and Security Applications of Evolutionary Computation

S.C. Upton, D. Goldberg

http://www-illigal.ge.uiuc.edu/msaec2004/

Almost since its inception, evolutionary computation has been applied to the solution of military problems. Since September 11, 2001, there has been increased interest within the military and security communities in novel techniques for solving challenging problems within their domains. The genesis of this interest lies in the fact that repeated attempts of using traditional techniques have left many important problems unsolved, and in some cases, not addressed. Additionally, new problems have emerged within the broad areas of the global war on terrorism, homeland security, and force protection that are difficult to tackle with conventional methods, since social, cultural and human behavioral factors tend to be at the heart of these new types of problems.

The purpose of the workshop is to introduce and discuss current and ongoing efforts in using genetic and evolutionary computation techniques in attacking military and security problems. These include, but are not limited to the following:

* genetic and evolutionary techniques in the design of military systems and sub-systems.

* genetic and evolutionary techiques for logistics and scheduling of military operations.

* genetics and evolutionary algorithms (GEAs) in strategic planning and tactical decision making.

* Multiobjective GEAs for examining tradeoffs in military, security, and counter-terrorism procedures;

* Automated discovery of tactics and procedures for site security, force protection, and consequence management;

* Genetics-based knowledge discovery and data mining of large databases used to recognize patterns of individual behavior;

* Co-evolutionary for simultaneous red-blue team strategic-tactical simulation and gaming.

The workshop invites completed or ongoing work in using evolutionary computation techniques for addressing these or any other application of genetic or evolutionary computation to military and security problems. This first workshop is intended to encourage communication between active researchers and practioners to better understand the current scope of efforts within this domain. The ultimate goal is to understand, discuss, and help set future directions for genetic and evolutionary computation in military and security problems.

Modularity, regularity and hierarchy in open-ended evolutionary computation.

H. Lipson, E. De Jong, J. Koza

http://www.mae.cornell.edu/lipson/gecco_modularity.htm

Scalability of open-ended evolutionary processes depends on their ability to exploit functional modularity, structural regularity and hierarchy. Functional modularity creates a structural separation of function that reduces the amount of coupling between internal and external behavior, allowing evolution to reuse modules as high-level building blocks. Structural regularity is the correlation of patterns within an individual, such as symmetry, repetition and self-similarity, allowing evolution to specify increasingly extensive structures while maintaining short description lengths. Hierarchy is the recursive composition of function and structure into increasingly larger and adapted units, allowing evolution to search efficiently increasingly complex spaces. This workshop will bring together researchers interested in these topics to discuss how principles of modularity, regularity and hierarchy can be applied in open-ended evolutionary computation.

Evolvability in Evolutionary Computation (EEC)

H. Suzuki, H. Sawai

http://www.his.atr.jp/%7Ehsuzuki/confs/2004_GECCO-WS-EEC.html

Evolvability is one of the most controversial issues in the studies on artificial evolutionary systems. Using various notions and definitions on evolvability, a number of researchers have examined different aspects of evolvability so far: population's adaptivity to a certain environment, trait for perpetual changes of genotypes in the population, system's potential ability to evolve functions or solutions, etc.
In a long-range term, the evolvability governs the dynamics and the final outcome of natural or artificial evolution, so investigating evolvability leads us to the evaluation and improvement of the design of artificial evolutionary systems. Having these notions in mind, this workshop focuses on evolvability studied in evolutionary computation (EC). As part of GECCO-2004, the workshop aims to bring together researchers interested in evolvability in GAs, GP, and so on, look back the previous achievements for evolvability in EC, and find out a clue or clues to extend the previous achievements towards the progress in our understanding of EC mechanisms and the enhancement of the evolvability.

Topics covered by the workshop are, but are not limited to:

* Genotype representation
* Coding problem
* Genotype-to-phenotype mapping
* Evolution of translation
* Variability of fitness landscape
* Measuring, observing, or enhancing evolvability
* Evolution of evolvability
* Biological basis for EC
* Symbiogenesis
* Epigenetic Inheritance
* Coevolution
* Mathematical Models for evolvability

Interactive Evolutionary Computing

Ian Parmee

http://www.ad-comtech.co.uk/Workshops.htm.

There is a history of research relating to interactive evolutionary computing which, in the main, relates to partial or complete human evaluation of the fitness of solutions generated from evolutionary search. This has generally been introduced where quantitative evaluation is difficult if not impossible to achieve. Examples of application include graphic arts and animation (Sims K ,1991; Sims K.,1991b); automotive design (Graf J., Banzhaf W.,1995); food engineering (Herdy M., 1997.) and database retrieval (Shiraki H., Saito H., 1996.) Such applications rely upon a human-centred, subjective evaluation of the fitness of a particular design, image, taste etc as opposed to an evaluation developed from some analytic model.

Partial human evaluation / interaction is also in evidence. For instance, user interaction relating to an evolutionary nurse scheduling system where a schedule model provides a quantitative evaluation of a solution. However, the model may not prove adequate in terms of changing requirements, qualitative aspects etc. In this case the user must add new constraints in order to generate solutions that are fully satisfactory (Inoue T., et al., 1999). In the pharmaceutical industry Computational Biology involves the modelling of biomolecular systems. Genetic algorithms (GA) can provide the search process for the identification of optimal biomolecule combinations. The process can be enhanced, however, by the user-introduction of new combinations as an elite solution into selected GA generations (Levine D. et al, 1997).

All the above applications utilise a major advantage of stochastic population-based search techniques. This relates to their capabilities as powerful search and exploration algorithms that can provide diverse, interesting and potentially competitive solutions to a wide range of problems. Parmee et al (1999, 2000, 2001, 2002) propose that such solutions can also provide information to the user which supports a better understanding of the problem domain whilst helping to identify best direction for future investigation. This perspective relates to human interaction when operating within ill-defined and uncertain decision-making environments in order to improve definition,
increase confidence and identify innovative / creative design
direction. The role here for evolutionary computation relates to exploration and the gathering of optimal information from simple conceptual models of the problem space. Such information supports model development by the user in an iterative, interactive EC environment where the first task is to evolve the problem space before attempting to solve the problem.

Optimization by Building and Using Probabilistic Models

M. Pelikan, K. Sastry, D. Thierens

http://www.cs.umsl.edu/~pelikan/obupm2004/

Genetic and evolutionary algorithms (GEAs) evolve a population of candidate solutions to a given optimization problem using two basic operators: (1) selection and (2) variation. Selection introduces a pressure toward high-quality solutions, whereas variation ensures exploration of the space of all potential solutions.

Two variation operators are common in current genetic and evolutionary computation (GEC): (1) crossover, and (2) mutation. Crossover creates new candidate solutions by combining bits and pieces of promising solutions, whereas mutation introduces slight perturbations to promising solutions to explore their immediate neighborhood. However, fixed, problem independent variation operators often fail to effectively exploit important features of high-quality solutions obtained by selection. One way to make variation operators more powerful and flexible is to replace traditional variation of GEAs by the following two steps:

1. Build a probabilistic model of the selected promising solutions, and

2. sample the built model to generate a new population of candidate solutions.

Algorithms based on this principle are called probabilistic model-building genetic algorithms (PMBGAs), estimation of distribution algorithms (EDAs), or iterated density-estimation evolutionary algorithms (IDEAs). The purpose of this workshop is to present and discuss

1. recent advances in PMBGAs,
2. new theoretical and empirical results,
3. applications of PMBGAs, and
4. promising directions for future PMBGA research.

International Workshop on Learning Classifier Systems (IWLCS2004)

W.Stolzmann, P.L. Lanzi, S.W.Wilson

No web page available yet.

IWLCS deals with current research on Learning Classifier Systems.

This will be the seventh IWLCS and the 4th to be held during GECCO. The workshop format will be the same as before, talks and discussions with a final discussion at the end of the workschop.
4. The names and full contact information (e-mail and postal addresses, fax, and telephone numbers) of the workshop organiser(s) and brief descriptions of their relevant expertise.

GECCO 2004 will have an LCS tutorial and an LCS track, so this workshop is interesting for all people who are interested in this tutorial and this track.

--CANCELED-- Learning, Adaptation, and Approximation in EC

J. Ocenasek, S. Mueller, S. Kern, N. Hansen, P. Koumoutsakos

The goal of this workshop is to provide a platform for the exchange of ideas between researchers investigating adaptivity in evolutionary algorithms and researchers investigating the development of optimization algorithms using concepts of approximation.

While in adaptation we are interested in learning distributions to search the parameter space efficiently, approximation aims at learning the objective function surface.

Grammatical Evolution

M. O'Neill, C. Ryan

http://www.grammatical-evolution.org/gews2004/index.html

Grammatical Evolution (GE) is an automatic programming system that can evolve programs in an arbitrary language from a binary string. GE adopts a genotype-phenotype mapping process taking as input a grammar that describes the syntax of the evolved program. In addition to the grammar, the search algorithm (the standard has been a variable-length genetic algorithm) is also a 'plug-in' component of the system.

The workshop will address all aspects of GE including foundations, extensions, analysis and applications.

Neutral Evolution in Evolutionary Computation

Tina Yu

http://www.improvise.ws/Workshop.htm

Kimura's Neutral Theory of Evolution is founded on the premise that most mutations at the molecular level in evolution are caused by random genetic drift rather than by natural selection. This contrasts to Darwin's Theory of Evolution which considers selection acting on advantageous mutations as the driving force of evolution. With a strong Darwinian influence, most EC systems adopt a selectionist's point of view to model evolution. It's only recently when neutrality is considered in EC systems. However, as the implementation differs, the performance results reported are different from one to the other. Currently, there is no consensus of the advantages/disadvantages of neutrality in EC. The purpose of this workshop is to discuss different views of neutrality and to improve our understanding of evolutionary search process under neutrality.

Regeneration and Learning in Developmental Systems (WORLDS)

J. F. Miller

http://www.elec.york.ac.uk/intsys/users/jfm7/worlds.htm

The workshop is concerned with constructing systems that autonomously grow using computational development and evolution.

Many biological organisms are multicellular. This means that complex genotype-phenotype mapping are being used. The complexity of living systems is orders of magnitude greater than man made systems. In the GEC community the dominant paradigm is Darwinian evolution. This is only one aspect of the evolution of living systems. We need to understand how to evolve small genotypes and through development and emergent interaction obtain evolved complex systems. The workshop welcomes innovative work in the exciting and relatively unexplored area.

Topics include:

- Design and evolution of developmental systems
- Evolution of complex systems (i.e. with potential unlimited numbers
of interacting parts)
- Relationship between development, self-repair and regeneration
- Evolvability of developmental systems
- Models of cellular processes that lead to multicellularity
- Genetic representations for developmental systems
- Exploitation of Emergence in Developmental Systems
- Autonomous learning in developmental systems
- Relationship between development and neural constructivism

Self-Organization in representations for evolutionary algorithms: Building complexity from simplicity

I. Garibay, G. Holifield, A. S. Wu

http://ivan.research.ucf.edu/SOEA.htm

The success of evolutionary algorithms in a wide range of otherwise intractable problems has promoted its use. As evolutionary algorithms are applied to increasingly difficult problems that require increasingly complex solutions, they face a number of problems: premature convergence to suboptimal solutions, stagnation of search in large search spaces, negative epistatic effects, disruption of large building blocks, among others. Natural evolutionn, on the other hand, seems to not have any problem evolving strikingly complex self-organized solutions. Self-organization is present in almost every level of natural evolution: gene regulation networks, proteins interaction networks, metabolic pathways, cellular organization, etc; but it is not usually present in evolutionary algorithms. Nature evolves instructions that produce organisms by a process of self-organization. Perhaps the self-organization of genotypic instructions into phenotypes is a key missing ingredient necessary for unleashing the evolution of complex and scalable solutions with emergent phenomena such as: scale-free-ness, adaptability, innovation, evolvability, and robustness. This workshop will focus on domain-independent methods for representing complex solutions with relatively simple self-organizable building blocks.

Topics of interest include (not limited to)

- Models of complexity building using self-organization
- Self-organized development: embryogenesis, growth
- Emergent behavior in representations
- Methods of fitness assignment for self-organized individuals (the price of non-programmability)
- Methods of design and evaluation of self-organizable building blocks
- Scalability of self-organizational processes to high complexities
- Self-organization theoretical approaches: complexity, chaos, synergetics, self-organized criticality, non-equilibrium thermodynamics, etc.
- Artificial self-organized systems
- NFL: what can we trade to get complexity and scalability in solutions?

This workshop seeks to bring together researchers from diverse problem domains to informally discuss issues related to the representation of complex solutions using self-organization of simple building blocks for evolutionary algorithms in particular, and the issue of building complexity from simplicity in general. We welcome technical papers describing completed or on-going research as well as position papers outlining current research issues, approaches or research agendas. We also welcome suggestions to panel discussions. Presentations will be short but will include extra time allocated for discussion. Preprints will be circulated by email prior to the meeting.

Graduate student workshop

Terry Riopka

http://www.kantbelievemyeyes.com/GraduateStudentWorkshop.html.

This single day workshop will involve approximately 12 selected students researching any aspect of Evolutionary Algorithms and presenting a a 15-20 minute synopisis of their current research to a mentor panel, other students and other selected participants. Each presentation will be followed by questions and discussion prompted by the mentor panel. The panel will consist of a rotating group of well known and established researchers in Evolutionary Computation. A limited number of other students will also be invited to attend the workshop where they will have an opportunity to join in discussions.

This format is intended to offer feedback from the panel to the presenters regarding their results, research methodology, future directions and presentation style. It should benefit other attendees in terms of learning about the work of others, engaging in technical discussions and meeting researchers with related interests. Workshops with approximately the same goals and format were held at previous GP/GECCO events and were strongly endorsed by both faculty and student participants.

The group of presenting students will be chosen by the panel with the intent of creating a diverse group of students working on a broad range of topic areas. You are an ideal candidate if your thesis topic has already been approved by your university and you have been working on your thesis for between 6 and 18 months. You are also a strong candidate if evolutionary computation has a role in an undergraduate project or thesis.

Importantly, even if you are not chosen to present, you will be considered for invitation to the workshop and you can expect to derive a lot of benefit from attending. Participation will be limited to preserve the discussion quality of the workshop but students who submit a paper will receive highest consideration. The papers submitted by students who participate in the workshop and/or presentation sessions will be printed in the GECCO Workshop Papers book. Awards will also be presented for best work, and best presentation.

Undergraduate student workshop

Mark M. Meysenburg

http://ist.doane.edu/meysenburg/gecco_ugws.htm

The workshop would serve as an opportunity for undergraduate students, and their faculty mentors, to present evolutionary computation work they have done for class projects or for more in-depth undergraduate research activities. Particulars of our proposed workshop are contained in the sections that follow.

We ran a very successful undergraduate workshop last year in Chicago, with six "formal" participants and presentations from several students who were not able to submit papers by the workshop deadline. Several panel members and teaching faculty members have inquired about having another workshop in Seattle.

Goals of the workshop would include:

* providing a forum allowing undergraduate students to put a "capstone" on their undergraduate research activities, through presentation of their work at an international conference;

* encouraging teaching faculty to think about undergraduate research opportunities for their students in the evolutionary computation field;

* preparing standout undergraduate students for graduate studies in the evolutionary computation field;

* encouraging more focus on education amongst GECCO participants;

* recruiting opportunities for faculty at advanced degree granting institutions; and

* sharing and networking amongst teaching faculty with students participating in undergraduate research.

Evolutionary Computation Theory

A. Wright, N. Richter

http://www.cs.umt.edu/u/wright/geccotheory/

This workshop will focus on genetic and evolutionary computation theory. Participants in the workshop will discuss strengths and weaknesses of different approaches to theory. We will hope to help researchers who are new to theory get started in the area and to help experienced researchers consider new directions.

The goal of evolutionary computation theory is to understand how the algorithms and methods of evolutionary computation work. We believe that this understanding is very important to guide more practically-oriented work in the field.

The format of the workshop will be the presentation of a limited number of papers interspersed with discussion. The paper presentation schedule will be flexible to allow plenty of time for discussion.

We invite papers on all areas of theory. We are interested in papers that give an overview of a particular approach to theory, or which compare the strengths and limitations of different approaches to theory. Papers that would help researchers who are new to theory get started in the area would be welcomed. A important criterion for inclusion will be the ability of a paper to stimulate discussion.

Biological Applications of Genetic and Evolutionary Computation (BioGEC)

http://chgr.mc.vanderbilt.edu/BioGEC/

The field of Genetic and Evolutionary Computation (GEC) has greatly benefited by borrowing ideas from the biological sciences. Recently, it has become clear that GEC can help solve biological problems, and thereby repay its debt.

The third annual workshop on Biological Applications of Genetic and Evolutionary Computation (BioGEC), organized in connection with the 2004 Genetic and Evolutionary Computation Conference (GECCO-2004) in Seattle, is intended to explore and critically evaluate the application of GEC to biological problems. Specifically, the goal is to bring biologists and computer scientists together to foster an exchange of ideas that will yield emergent properties that will move the field forward in unpredictable ways.

In order to facilitate interaction and discussion, the workshop invites papers in the form of commentaries, essays, perspectives, surveys, tutorials, and reviews that focus on ideas for discussion rather than specific research results. Investigators interested in presenting research results are encouraged to submit their papers to the GECCO track on biological applications. Questions that might be addressed in a paper include (but are not limited to):

1) What biological problems are GEC methods well-suited for?
2) What biological problems are GEC methods not well-suited for?
3) Which of the many GEC methods should be used for a specific
biological problem?
4) What are the successes and failures of GEC for a specific
biological problem?
5) What impact has GEC had on biology/bioinformatics?
6) Should all biologists/bioinformaticists be using GEC?
7) What is the future of GEC for solving biological problems?
8) What GEC software tools are available for use by
biologists/bioinformaticists? 9) What unanswered questions in GEC
are relevant to solving biological problems?