Evolutionary Computation Technical Committee
Task Force on
Evolutionary Computation in
Dynamic and Uncertain Environments (ECiDUE)
Chair:
Members:
Dirk Arnold, Dalhousie University, Canada
Juergen Branke, University of Karlsruhe, Germany
Andries Engelbrecht, University of Pretoria, South Africa
Ernesto Costa, University of Coimbra, Portugal
Sima Etaner-Uyar, Istanbul Technical University, Turkey
Chi-Keong Goh, Data Storage Institute, Singapore
Xiaodong Li, RMIT University, Australia
Ferrante Neri, University of Jyvaskyla, Finnland
Yew Soon Ong, Nanyang Technological University, Singapore
Ke Tang,University of Science and Technology of China, China
Renato Tinos, Universidade de Sao Paulo (USP), Brazil
Khaled Rasheed, The University of Georgia, USA
Shengxiang Yang, University of Leicester, UK
Target and Motivation
The primary target of the task Force is to promote research on
evolutionary computation in dynamic and uncertain environments.
This is an emerging area in evolutionary computation,
which covers the following different but closely related
topics:
- Evolutionary computation (optimization) with noisy fitness evaluations.
Noise in fitness evaluations may result from many different sources such
as sensory measurement errors or numerical instabilities in simulation.
Basic strategies to handle noisy fitness functions include
population sizing, averaging by re-sampling, or changing
the selection criteria.
- Evolutionary computation (optimization) with approximate or imprecise
fitness (quality) evaluations. The primary motivation to
use computationally efficient but
imprecise fitness evaluations is that the original fitness
function is too expensive or does not exist.
In comparison to noisy fitness functions,
uncertainties introduced by approximate fitness evaluations are biased
and cannot be reduced by re-sampling the approximate fitness function.
The most popular method to obtain computationally efficient
fitness evaluations is to construct a meta-model (surrogate) based
on available data. Ad hoc methods such as fitness inheritance,
fitness imitation or fitness assignment can also be employed.
- Evolutionary optimization where the design variables or the environmental
parameters are subject to stochastic or deterministic changes.
It is very common that a system to be optimized is expected to perform
satisfactorily even when the design variables or the environmental
parameters change within a certain range, or the system has to work on
more than one pre-determined nominal point. This issue is often known as
the search for robust optimal solutions.
- Evolutionary computation with a time-varying fitness function.
In other words, the optimum of the system is changing with time, which
requires a repeated re-optimization or even continuous tracking of
the optimum. As a matter of fact, in multi-objective evolutionary
computation and co-evolutionary systems, the fitness landscape changes
over time.
Organized Journal Special Issues, Conference Special Sessions and
Workshops
- Special Session on Evolutionary Computation in Dynamic and Uncertain
Environments at CEC 2009, Trondheim, Norway, 18-21 May, 2009
- The 6th European Workshop on Evolutionary Algorithms in Stochastic and
Dynamic Environments (EvoSOTC 2009), Tübingen, Germany, 15-17 April, 2009.
- Special Session on Evolutionary Computation in Uncertain Environments at
WCCI 2008, Hong Kong, 1-6 June, 2008
- The 5th European Workshop on Evolutionary Algorithms in Stochastic and
Dynamic Environments (EvoSTOC 2008), Napoli, Italy, 26-28 March, 2008.
- Special Session on Evolutionary Computation in Dynamic and Uncertain
Environments at CEC 2007, Singapore, 25-28 September, 2007
- The 4th European Workshop on Evolutionary Algorithms in Stochastic and
Dynamic Environments (EvoSOTC 2007), Valencia, Spain, 11-13 April, 2007
- Special Issue on Evolutionary Optimization in the Presence of
Uncertainties, IEEE Transactions on Evolutionary Computation,
10(6), 2006
- Special Issue on ECiDUE, Genetic Programming and Evolvable
Machines, 7(4), 2006
- Special Session on Evolutionary Computation in Dynamic and Uncertain
Environment, WCCI 2006, Vancouver, Canada, 16-21 July, 2006
- The 3rd European Workshop on Evolutionary Algorithms in Stochastic and
Dynamic Environments (EvoSOTC 2006), Budapest, Hungary, 10-12 April, 2006
- Special Session on Evolutionary Computation in Dynamic and Uncertain
Environments, CEC 2005, Edinburgh, UK, 2-5 September, 2005
- The 2nd European Workshop on Evolutionary Algorithms in Stochastic and
Dynamic Environments (EvoSOTC 2005), Lausanne, Switzerland, March 30-April 1, 2005
- Special Session on Evolutionary Optimization in Dynamic Environments,
CEC 2004, Portland, USA, 19-23 June 2004
- Special Session on Evolution and approximation in design optimization,
CEC 2004, Portland, USA, 19-23 June 2004
- The 1st European Workshop on Evolutionary Algorithms in Stochastic and
Dynamic Environments (EvoSOTC 2004), Coimbra, Portugal, 5-7 April 2004.
Resources
-
Benchmark problems used for competition at CEC 2009
- An edited book on Evolutionary Computation in Dynamic and
Uncertain Environments edited by S. Yang, Y.S. Ong and Y. Jin, Springer, 2007
- A tutorial on
Fitness Approximation in Evolutionary Computation by Yaochu Jin and
Khaled Rasheed on Genetic and Evolutionary Computation Conference,
June 26, Washington D.C., 2005
- A survey paper on evolutionary optimization in uncertain environments by Yaochu Jin and Juergen Branke.
Published in IEEE Transactions on Evolutionary Computation,
9(3):303-317, 2005
- A tutorial on
Evolutionary computation in dynamic and uncertain
environments by Yaochu Jin on CEC'04, Portland, July 2004
- A paper survey on
fitness approximation in evolutionary
computation by Yaochu Jin. Published in Soft Computing, 9(1):3-12, 2005
- A bibliography on
Evolutionary Computation with Approximate Fitness Functions
- A bibliography on
Evolutionary Optimization in/for Noisy Environments
- A bibliography on Evolutionary Algorithms for Dynamic Optimization Problems
Note: The copyright of the unpublished slides available on this homepage is
reserved by the author. Use all or part of the materials for any purpose
other than personal use, such as lecture handouts, is allowed but should
be properly acknowledged.
Last updated in June, 2009.