Comparing approximate reasoning and probabilistic reasoning using the Dempster–Shafer framework

Abstract

We investigate the problem of inferring information about the value of a variable V from its relationship with another variable U and information about U. We consider two approaches, one using the fuzzy set based theory of approximate reasoning and the other using probabilistic reasoning. Both of these approaches allow the inclusion of imprecise granular type information. The inferred values from each of these methods are then represented using a Dempster–Shafer belief structure. We then compare these values and show an underling unity between these two approaches.