Context Recognition of Noisy Data Employing

Complementary Imprecise Decision Tables

D.J. Poling

Department of Computer Science

Clemson University

Clemson, SC 29634-1906 USA

[email protected]

February 12, 1993

Abstract

Employment of complementary imprecise decision tables within a hybrid system structure allows significant, automatic pattern recognition of noisy data in context. This approach greatly enhances the probability of accuracy when dealing with noisy data.

In this paper an artificial neural network is used to cluster the input data. This classification information is then input to an imprecise decision table which works in close conjunction with an expert system. The expert system derives information pertaining to context patterns and feeds this information into another imprecise decision table. This later table then expresses an opinion as to the confidence of the pattern recognition based not only upon the classification, but also upon previous and commensurate patterns.
KEYWORDS: fuzzy logic, pattern recognition, decision table, imprecise decision table For a copy of this paper, please contact the author.