Special Session on Performance Assessment of Multi-Objective Optimization Algorithms @ CEC-07, Singapore, 25-28 September 2007


For recent updates click here



Optimization for multiple conflicting objectives results in more than one optimal solutions (known as Pareto-optimal solutions). Although one of these solutions is to be chosen at the end, the recent trend in evolutionary and classical multi-objective optimization studies have focused on approximating the set of Pareto-optimal solutions. It is then believed that such a set of solutions will collectively provide a good insight to the different trade-off regions on the Pareto-optimal front, thereby aiding a better and more confident decision making at the end. However, which type of approximation of the Pareto-optimal set is sought strongly depends on the decision maker; here, various aspects such as convergence to the Pareto-optimal front and maintenance of diversity among the obtained solutions come into play. Thus, to assess the performance of such optimization algorithms, the decision maker's preferences need to be taken into account.


Evolutionary multi-objective optimization (EMO) methodologies were suggested in the early Nineties for this task, and since then a number of performance assessment methods have been suggested. Most of the existing simulation studies comparing different EMO methodologies are based on specific performance measures. After more than 10 years of research and development of efficient EMO algorithms, we realize that it is time to evaluate the existing EMO and classical generating methodologies on carefully chosen test problems and practical problems which are scalable with respect to the objectives and the decision variables. The goal is to consider different types of preferences, e.g., formalized in terms of appropriate performance measures, so as to bring out the essential features needed in an algorithm to efficiently solve multi-objective optimization problems depending on the decision maker's preferences. The comparisons will be made for a limited number of overall evaluations, so that the existing or new algorithms can be evaluated for different functional abilities:


i) to meet well specified preferences (convergence to Pareto front, diversity, objective values, etc.)

ii) to scale well on many objectives, and

iii) to scale well on many variables.


Following the successful organizations of two other special sessions on unconstrained and constrained single-objective optimization (held in CEC-05 and scheduled in CEC-06, respectively), during CEC-07, we shall organize this special session on multi-objective optimization algorithms. We shall develop a set of scalable test problems providing different kinds of complexities, a set of different, commonly considered preference types following the recent literature, a careful plan for execution of simulations and a presentation format, so interested participants can put to test their already published or modified algorithms. We hope to publish the edited volume as Springer's Lecture Notes in Computer Science after the conference.


With this background, we now invite you give your feedbacks / suggestions on developing a test suite with appropriate evaluation methods and would like to know if you would be willing to participate in this exercise. Any sort of search engine is allowed, including hybrids with mathematical programming techniques as well as different metaheuristics. Please could you kindly send an email to all the organizers with the following details?






Methods to be used:  (a) EMO  (b) Classical Generating Method (c) Hybrid

If you know of researchers who might be interested in making contribution(s), please

kindly provide names/email addresses. Thank you.



We hope to have the test functions available by the end of October 2006 from http://www.ntu.edu.sg/home/EPNSugan. Paper submission deadline is 15 March 2007.


Thank you

Special Session Organizers:

Prof. Kalyanmoy Deb                      (deb@iitk.ac.in)

A/Prof. P. N. Suganthan                  (epnsugan@ntu.edu.sg)

A/Prof. Eckart Zitzler                       (zitzler@tik.ee.ethz.ch)