 |
||
   |
          |
|
       |
|
Tutorials held prior to the congress serve to update participants on the advancements in the field of evolutionary computation topics in recent years. The IEEE CEC 2007 will feature a number of pre-congress tutorials that are related to the technical scope of the conference on 25 September 2007: Tutorials Session 1: (8.15am to 11am) Lunch Break: (11am to 12noon) Tutorials Session 2: (12noon to 2.45pm) Tea Break: (2.45pm to 3.15pm) Tutorials Session 3: (3.15pm to 6pm) |
| Abstract | ||
| Tut. 1. Title: Evolutionary Computation: A Unified Approach Organizer: Kenneth A. De Jong Venue: Room Enterprise The field of Evolutionary Computation (EC) has experienced tremendous growth over the past 15 years, resulting in a wide variety of evolutionary algorithms and applications. The result poses an interesting dilemma for many practitioners in the sense that, with such a wide variety of algorithms and approaches, it is often hard to se the relationships between them, assess strengths and weaknesses, and make good choices for new application areas. This tutorial is intended to give an overview of EC via a general framework that can help compare and contrast approaches, encourage crossbreeding, and facilitate intelligent design choices. The use of this framework is then illustrated by showing how traditional EAs can be compared and contrasted with it, and how new EAs can be effectively designed using it. Finally, the framework is used to identify some important open issues that need further research.
|
||
Tut. 3. Title: Evolutionary Computation in Finance and Economics
Organizer: Edward Tsang Venue: Room Mercury Computing has changed many aspects of our daily life. It certainly has changed the face of finance and economics research. Advances in both hardware and software allow us to study finance and economic in ways that were previously impossible. For example, advances in evolutionary computation allows us to discover forecasting rules, bargaining strategies and economic models; advances in optimization allows us to search for portfolios with maximum return and minimum risk. The use of artificial markets enable us to understand better some of the fundamental concepts in economics, such as rationality and the efficient market hypothesis. In this tutorial, I shall give an overview of research in evolutionary computation in finance and economics. Top
|
||
| Tut. 4. Title: From Evolving Single Neural Networks to Evolving Ensembles Organizer: Xin Yao Venue: Room Minto Designing compact neural networks that generalise well is often very difficult and requires much domain knowledge about the problem. Simulated evolution can be used to search for a near optimal neural network automatically. This tutorial first introduces an evolutionary system for designing neural networks, where behavioural rather than genetic evolution is emphasised. Parsimony is encouraged through the sequential application of mutation operators. Behavioural link between parents and offspring is maintained. Some experimental results will be given. The tutorial then argues that learning is different from optimisation in practice. The way we evolved neural networks did not exploit the useful population information fully. It is shown, through experiments, how simple techniques can be used to improve neural network's generalisation ability by combining different individuals in a population. Since we are going to combine different individual neural networks, the question now is how to design and train those individuals so that they can be combined more effectively. This tutorial describes a new learning algorithm which trains individual neural networks simultaneously so that they will cooperate with each other when combined. OUTLINE: 1. Introduction to Evolutionary Computation
|
||
Tut. 6. Title: Evolving Soccer Teams for RoboCup Simulation
Organizer: Tomoharu Nakashima Venue: Room Enterprise RoboCup is an international project that promotes various research fields such as robotics and artificial intelligence. The ultimate aim of RoboCup soccer is to beat against the human world champion team by the year 2050. The environment of RoboCup soccer provides us with many challenging problems such as real-time information processing, noisy data handling, and cooperation between soccer players. Among five leagues in RoboCup soccer, simulation league is the most active research domain in terms of the development of computational intelligence techniques. In the first part of the tutorial we give a general introduction of RoboCupand RoboCup soccer. Specifically the soccer simulation league is more focused on than the other leagues. The second part of the tutorial shows an evolutionary computation method for evolving soccer team strategies. A team strategy is encoded as an integer string that represents a set of action rules of ten players. The task of the evolutionary computation method is to find the optimal set of action rules. Various techniques for improving the performance of the evolutionary computation method are also illustrated. For example, instability in the game results obtained from the iterations of the same match-up is resolved by modifying the fitness function in the evolutionary computation method. The tutorial is intended to give an idea of how to implement computational intelligence techniques especially evolutionary computation in developing RoboCup soccer simulation teams. A simple implementation of a soccer team using a base set of source codes will be also demonstrated.
|
||
Tut. 7. Evolutionary Computation in Bioinformatics Application of evolutionary computation to problems in computational
biology and bioinformatics have steadily increased in recent years.
This is likely due to the demonstrated utility of evolutionary algorithms
for searching vast numbers of possible solutions and as an optimization
method for pattern recognition models. This tutorial will provide an
introduction to a wide assortment of biological problems
relevant to the computer scientist as well as an overview of the leading
data repositories (EMBL, GenBank, DDBJ, Swiss-Prot, PDB, and
others) typically utilized when working on computational biology and
bioinformatics research. Applications of evolutionary computation will be illustrated, in terms of both broad possibility and specific applications.
The application section of the tutorial will be of interest to both
biologists and computer scientists. The tutorial will conclude by
identifying open opportunities for further exploration in this regard.
|
||
Tut. 8. Title: Evolutionary Multi-Criterion Optimization (EMO): Fundamentals,
State-of-the-art Methodologies and Future Challenges Many real-world search and optimization problems are naturally posed
as non-linear programming problems having multiple conflicting objectives. Due to
lack of suitable solution techniques, such problems are usually artificially
converted into a single-objective problem and solved. The
difficulty arises because multi-objective optimization problems give
rise to a set of Pareto-optimal solutions, each corresponding to a
certain trade-off among the objectives. It then becomes important to
find not just one Pareto-optimal solution but as many of them as possible.
Classical methods are found to be not efficient because they require
repetitive applications to find multiple Pareto-optimal solutions and
in some occasions repetitive applications do not guarantee finding
distinct Pareto-optimal solutions. The population approach of
evolutionary algorithms (EAs) allows an
efficient way to find multiple Pareto-optimal solutions simultaneously in
a single simulation run.
|
||
Tut. 9. Title: Adaptive Business Intelligence--Evolutionary Computation for Real World Problems
|
||
| Tut. 11. Title: Genetic Programming Practice and Theory Organizer: Riccardo Poli Venue: Room Mercury Genetic programming (GP) is an evolutionary technique for getting
computers to automatically solve problems without having to tell them
explicitly how to do it. Since its inception genetic programming has This tutorial includes two parts. In the first part I will introduce the basics of GP practice, briefly explaining GP representations, operators and search algorithm, and showing examples of real runs. In the second part of the tutorial, I will then concentrate on explaining how and why GP works. This will done by first characterising GP's search space (the space of all possible programs) and then by explaining the way in which GP explores such a space. Despite its technical contents, the tutorial will require limited
mathematical background.
|
||
Tut. 12. Title: Particle Swarm Optimization and Differential Evolution Since the first publication of Particle Swarm Optimization (PSO) in 1995, the number of research papers on PSO and the number of researchers in PSO have exploded. Many variations of the PSO have been developed to improve the performance, studies have been done to understand the dynamics of particles, and adaptations have been developed to apply the PSO to different optimization problem types. DE algorithm, proposed by Storn and Price in 1995, is a simple but powerful population-based stochastic search technique for solving global optimization problems. Its efficiency has been successfully demonstrated in many application fields. DE’s self-organizing scheme takes the difference vector of two randomly chosen population vectors to perturb an existing vector. This tutorial will provide the attendee with an overview of PSO and DE literature, and will show how the different PSO/DE algorithms can easily be implemented using an open source CI library, called Cilib. This will provide the attendees with hands-on experience in using and coding PSO/DE algorithms.
The above topics will be illustrated during the tutorial using the open source library, Cilib.
|
|