MATHEMATICS COLLOQUIUM

 WHEN:
On Thursday, September 20, 2001 at 2.30PM

WHERE:
Durham Science Center, Room 255

WHAT:

Zhenyuan Wang

University of Nebraska at Omaha

will give a talk on

Part I: Nonadditive set functions and relevant nonlinear integrals

Abstract:
A basic assumption in using linear multiregression to depict how the objective attribute depends on the predictive attributes in a database is no interaction existing among the latter toward the former, that is, the joint contribution toward the objective attribute from a set of predictive attributes is just the sum of contributions from each individual predictive attribute in the set. However, this linear model is not a good approximation in many real databases where the interaction among attributes cannot be ignored. In such a case, a nonadditive set function may be used to describe the interaction, and a proper nonlinear integral with respect to the nonadditive set function should be adopted as an aggregation tool to replace the classical Lebesque-like integral.