On the meta-knowledge choquet integral and related models

Abstract

Choquet integral and multistep Choquet integrals have been used in recent years as models for decision making and information aggregation. Such models can be used to fuse information when information sources are not independent. A basic property of such models is that their output is monotonically increasing with respect to inputs. In this article we study two alternative models built on the basis of such Choquet integrals. The motivation is, on the one hand, to study the modeling capabilities of such operators and, on the other, to build models that are capable of approximating any arbitrary functions (not only monotonic ones). In this article we describe and study two models that are universal approximators. © 2005 Wiley Periodicals, Inc. Int J Int Syst 20: 1017–1036, 2005.