| 摘要 |
PROBLEM TO BE SOLVED: To provide a multiobjective optimization apparatus, method, and program for making it possible to obtain a diversified Pareto-optimum individual in a short time even if an optimization target has uncertainty. SOLUTION: A multiobjective evolutionary algorithm unit 2 feeds a set of parameters of an individual to a search history storage device 31 in a fitness estimating unit 3 and to an optimization target 6. The optimization target 6 outputs a set of sampled values of fitnesses on the basis of the set of parameters of the individual. The search history storage device 31 stores the set of parameters of the individual and a set of sampled values as a search history. A fitness estimating module 30 computes a set of estimated values of true fitnesses on the basis of the search history stored in the search history storage device for output to the multiobjective evolutionary algorithm unit 2. The multiobjective evolutionary algorithm unit 2 determines a Pareto-optimal population in accordance with a genetic algorithm on the basis of a plurality of sets of estimated values. COPYRIGHT: (C)2006,JPO&NCIPI |
