Determining nonnegative monotone set functions based on Sugeno's integral: an application of genetic algorithms

Abstract

Regarding the set of all information sources as the universe of discourse, we used a nonnegative monotone set function defined on its power set to describe the importance of each individual information source and their varied combinations. Such a set function is called an importance measure or a fuzzy measure. The Sugeno integral with respect to the nonnegative monotone set function possesses many desired properties, such as the fuzzy linearity when the set function is fuzzy additive, and can be adopted as an aggregation means in information fusion. In real problems, viewing the Sugeno integral as a multi-input single-output system, we use a genetic algorithm to determine the values of the importance measure from the input–output data of the system.