2006 IEEE Congress on Evolutionary Computation

Sheraton Vancouver Wall Centre
Vancouver, British Columbia, Canada
16-21 July 2006

Special Session on Constrained Real Parameter Optimization


Most optimization problems have constraints of different types (e.g., physical, time, geometric, etc.) which modify the shape of the search space. During the last few years, a wide variety of metaheuristics have been designed and applied to solve constrained optimization problems. Evolutionary algorithms and most other metaheuristics, when used for optimization, naturally operate as unconstrained search techniques. Therefore, they require an additional mechanism to incorporate constraints into their fitness function.

 Historically, the most common approach to incorporate constraints (both in evolutionary algorithms and in mathematical programming) is the penalty functions, which were originally proposed in the 1940s and later expanded by many researchers. Penalty functions have, in general, several limitations. Particularly, they are not a very good choice when trying to solve problem in which the optimum lies in the boundary between the feasible and the infeasible regions or when the feasible region is disjoint. Additionally, penalty functions require a careful fine-tuning to determine the most appropriate penalty factors to be used with our metaheuristics.

 In order to overcome the limitations of penalty functions approach, researchers have proposed a number of diverse approaches to handle constraints such as fitness approximation in constrained optimization, incorporation of knowledge such as cultural approaches in constrained optimization and so on. Additionally, the analysis of the role of the search engine has also become an interesting research topic in the last few years. For example, evolution strategies (ES), evolutionary programming (EP), differential evolution (DE) and particle swarm optimization (PSO) have been found advantageous by some researchers over other metaheuristics such as the binary genetic algorithms (GA).

 Despite the existence constrained optimization test suites, there is an obvious need to upgrade the current test suites by considering the types of constraints (equality, inequality, linear, nonlinear, dimensionality, active, etc.), types of objective functions (linear, quadratic, nonlinear, multimodality, separability, etc.), connectivity, relative size of feasible region and so on. In addition, it would be beneficial to evaluate and, if necessary, develop novel performance measures to deal with the diverse characteristics of the constrained optimization problems. We plan to present an extended test suite and standardized evaluation measures for researchers to test their algorithms till the CEC'2006 submission deadline in late January 2006. Along with the papers, we would also optionally like participants to submit their codes and/or executables and we shall put it up on a web-site for anyone to try out. The submitted papers will be peer-reviewed by other authors and reviewers and selected authors will be invited to present their results during CEC-06. Later, we plan to put together an edited volume with more details, so that effective algorithms will be available in one volume with comparison results based on identical criterion and on identical test problems. We hope this exercise will be helpful for other researchers interested in this field and may generate new ideas for progressing the research in this area. We hope to publish the edited volume as Springer's Lecture Notes in Computer Science after the conference.

The 2006 IEEE Congress on Evolutionary Computation is part of the 2006 IEEE World Congress on Computational Intelligence.



Important dates:

Special Session Organizers:

Paper Submission: [Enter]

Author's Kit: [Enter] New!



Updates:

*if there is any file you can not open, please right click and choose "save as".

Mar 21,2006 New!

"Modifications for the final version" is added, please modify your final version according to the instruction. [download]

Best known solutions are updated according to the solutions provided by the participants. [bestknown_new.m] [bestknown_new.doc]

"Problem Definitions and Evaluation Criteria for the CEC 2006 Special Session on Constrained Real-Parameter Optimization" is updated.

Feb 4,2006

Some typing errors in g17 in "Problem Definitions and Evaluation Criteria for the CEC 2006 Special Session on Constrained Real-Parameter Optimization.pdf" is corrected.

Sorting method for the final results is added. If you used a different sorting method, please modify it now or in the final version.

Dec 20, 2005

A small error in Table 3 in "Problem Definitions and Evaluation Criteria for the CEC 2006 Special Session on Constrained Real-Parameter Optimization.pdf" is corrected.

Dec 14, 2005

A study using the MATLAB Optimization Toolbox is updated.

Dec 8, 2005   

"Problem Definitions and Evaluation Criteria" is updated and some errors are corrected. Please download the new one.

Please notice the best known solutions provided in the file are updated and bestknown.m is updated.

Dec 5, 2005   

c to java converter is found not work properly and the provided java code is not correct. New code will be provided later. Sorry for any caused inconvenience.

A test data set used for testing the correctness of the code is provided. [test data]



Downloads (last updated on 21/Mar/2006):

Modifications for the final version [pdf] New!

Please keep the variables within the provided search bounds. If you search out of the search bounds for g02, g14, g21 and g22, you may get -Inf or NaN. Please note that the search range for g02 and g14 is 0<xi<=10, not 0<=xi<=10

Problem Definitions and Evaluation Criteria for the CEC 2006 Special Session on Constrained Real-Parameter Optimization [pdf]

Code (could be employed by C/C++/C#,Matlab and Java) [Enter] (last updated on 11/Dec/2005)

Results Format [Latex] [Word]

*Sorting method for the final results :

1. Sort feasible solutions in front of infeasible solutions;

2. Sort feasible solutions according to their function errors f(x)-f(x*)

3. Sort infeasible solutions according to their mean value of the violations of all constraints.




If you face any difficulties, please inform  epnsugan@ntu.edu.sg