Genetic algorithms for determining fuzzy measures from data

Article type: Research Article

Authors: Wang, Wei | Wang, Zhenyuan | Klir, George J.

Affiliations: Department of Systems Science & Industrial Engineering, Thomas J. Watson School of Engineering and Applied Science, Binghamton University-SUNY, Binghamton, NY 13902-6000, USA

Abstract: A synthetic evaluation of a given object in terms of multiple factors that contribute to some feature of the object (quality, performance, etc.) may be regarded as a system with multiple inputs and one output. Traditionally, the output is expressed as the weighted average of the inputs. Unfortunately, this method is severely limited as it cannot capture any inherent relation among the factors involved. This limitation can be overcome by using the Choquet integral or the fuzzy integral with respect to a fuzzy measure that captures the relation among the factors. The crux of this method is to determine the right fuzzy measure. In this paper, we describe an efficient genetic algorithm for constructing a suitable fuzzy measure from relevant input–output data. This algorithm has a broad applicability in various problem areas, such as decision making, cluster analysis, pattern recognition, image and speech processing, and expert systems.

Journal: Journal of Intelligent and Fuzzy Systems, vol. 6, no. 2, pp. 171-183, 1998