Premium
Minimization of uncertainty for ordered weighted average
Author(s) -
Vergara Victor M.,
Xia Shan
Publication year - 2010
Publication title -
international journal of intelligent systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.291
H-Index - 87
eISSN - 1098-111X
pISSN - 0884-8173
DOI - 10.1002/int.20422
Subject(s) - minification , entropy (arrow of time) , variance (accounting) , weighted arithmetic mean , mathematics , similarity (geometry) , mathematical optimization , computer science , information fusion , fusion , data mining , algorithm , artificial intelligence , statistics , linguistics , physics , philosophy , accounting , quantum mechanics , business , image (mathematics)
Abstract Existing ordered weighted average (OWA) characterization methods maximize similarity among information sources by seeking maximal weights entropy or by minimizing weights variance. These methods are based solely on the weights, and the uncertainties of input information sources are ignored. However, the purpose of information fusion is to decrease uncertainty and improve data quality. Following this objective, this work proposes a new method to calculate the OWA weights based on the minimization of the aggregated uncertainty. The resulting aggregated value is the most precise, in the sense that any other combination of weights produces larger uncertainty. © 2010 Wiley Periodicals, Inc.