An Evolutionary Algorithm Guided by Preferences Elicited According to the ELECTRE TRI Method Principles
The resolution of a multi-objective optimization problem involves not just a search and computation phase, capable of providing a representative sample of the Pareto-optimal front, but also a decision support process to aid the Decision Maker (DM) to progress in the learning of the trade-offs at stake in different regions of the search space. This is accomplished by integrating in the search process the DM’s preferences to guide the search and limit both the cognitive effort, in assessing Pareto-optimal solutions with distinct characteristics, and the computational effort, by reducing the scope of the search according to the preferences expressed by the DM. The introduction of meaningful preference expression parameters used in the ELECTRE TRI method for sorting problems in the framework of an evolutionary algorithm is proposed. Illustrative results in an operational planning problem in electricity networks are reported.
KeywordsEvolutionary algorithm ELECTRE TRI handling preferences adaptive algorithms
Unable to display preview. Download preview PDF.
- 2.Coello, C.A.C., Veldhuizen, D.V., Lamont, G.B.: Evolutionary Algorithms for Solving Multiobjective Problems. Kluwer Academic Publishers, New York (2002)Google Scholar
- 4.Horn, J.: Multicriterion decision making. In: Bäck, T., Fogel, D., Michalewicz, Z. (eds.) Handbook of Evolutionary Computation, vol. 1, pp. F1.9:1–F1.9:15.5. Oxford Univ. Press, Oxford (1997)Google Scholar
- 5.Coello, C.A.C.: Handling Preferences in Evolutionary Multiobjective Optimization: A Survey. In: Proc. of the 2000 Congress on Evo. Computation, pp. 30–37 (2000)Google Scholar