Identification of λ-fuzzy measures using sampling design and genetic algorithms

The data collection and identification of λ-fuzzy measures are fundamental tasks to be capable of using them in practice. Although several studies have been made on λ-fuzzy measure identification, their methods require the knowledge of subjective estimates for all subsets of an attribute set, or the corresponding computation process is rather complicated. Since little attention has been given to the issue of data collection, this study designed a sampling procedure of attribute subsets to capture the most useful information and thus reduce the original information demand in solving λ-fuzzy measures. Then, this study developed an identification procedure for λ-fuzzy measures using genetic algorithms. Several experiments were implemented to show the applicability and feasibility of the proposed methods.