Evolutionary Multi-Criterion Optimization - Fonseca Carlos M.

Multi-criterionoptimizationreferstooptimizationproblemswithtwoormore- jectives expressing con?icting goals that are formulated within a mathematical programming framework. The problems addressed may involve linear or nonl- ear objective functions and/or constraints, continuous or discrete variables, and may or may not be a?ected by uncertainty in the data. This branch of multiple criteria decision making (MCDM) ?nds application in numerous domains: en- neering design, health, transportation,telecommunications, bioinformatics, etc. The concept of a unique optimal solution does not apply as soon as multiple objectives are optimized simultaneously. The models and methods introduced in multi-criterion optimization deal with the concept of a set of e?cient (also called Pareto optimal) solutions. E?cient solutions imply trade-o?s between the di?erentcriteria. Thecomputationofthee?cientsolutionsetmaybehardwhen the size of the problem is large, when the problem is computationally complex, when the data are not crisp. It is then often impossible to guarantee the com- tation of exact solutions. In that case, approximate solutions, i. e. , sub-optimal solutionscomputedwithlimitedandcontrolledresources,suchasavailabletime, are of interest. This is the domain of multi-objective metaheuristics, of which evolutionary multi-criterion optimization (EMO) is de?nitely the most pro- nent representative. The success of EMO is due to the simplicity of its concepts and the generality of its methods, and is clearly expressed by the many impr- sive success stories reported in the literature. Research activities in EMO have boomed since the mid-1990s. Three g- erations of work are identi?able throughout the years.

EAN: 9783642010194