Employing A Neural Network For Fuzzy Linear Regression

D.J. Poling
Department of Computer Science
Clemson University
Clemson, SC 29634-1906 USA
[email protected]

Douglas Lynch
Department of Computer Science
Florida State University
Tallahassee, FL 32304 USA
[email protected]

February 12, 1993

Abstract

Fuzzy linear regression is a method of forecasting in an uncertain environment. Unlike regular (crisp) linear regression, which assumes the deviation between the estimated and actual values is due to observation errors, fuzzy linear regression assumes this deviation is due to the fuzziness of the parameters.

This paper presents a use of a neural network for fuzzy linear regression. This network allows the fuzzy regression to be continually optimized after the initial regression has been completed.
Additionally, this paper discusses some of the areas for additional research.

For a copy of this paper, please contact the author.