®®®® SIIA Público

Título del libro: Genetic And Evolutionary Computation Conference, Gecco'11
Título del capítulo: A RankMOEA to approximate the Pareto Front of a Dynamic Principal-Agent model

Autores UNAM:
JUAN ARTURO HERRERA ORTIZ; KATYA RODRIGUEZ VAZQUEZ;
Autores externos:

Idioma:

Año de publicación:
2011
Palabras clave:

Evolutionary Algorithms; Multi-objective optimization


Resumen:

In this paper, a new Multi-Objective Evolutionary Algorithm (MOEA) named RankMOEA is proposed. Innovative niching and ranking-mutation procedures which avoid the need of parameters definition are involved; such procedures outperform traditional diversity-preservation mechanisms under spread-hardness situations. RankMOEA performance is compared with those of other state of the art MOEAs: MOGA, NSGA-II and SPEA2, showing remarkable improvements. RankMOEA is also applied to approximate the Pareto Front of a Dynamic Principal-Agent model with Discrete Actions posed in a Multi-Objective Optimization framework allowing to consider more powerful assumptions than those used in the traditional single-objective optimization approach. Within this new framework a set of feasible contracts is described, while others similar studies only focus on one single contract. The results achieved with RankMOEA show better spread and minor error than those obtained by already mentioned MOEAs, allowing to perform better economic analysis in the contracts trade-off surface. Copyright 2011 ACM.


Entidades citadas de la UNAM: