On a Reasoning Method Considering Uncertainty with Fuzziness
Rumiko AZUMA, Hayao MIYAGI
In the field of Artificial Intelligence, there are many studies of reasoning method with uncertainty. Recently, research about the reasoning method from observed fact with uncertainty has been developed, and also the reasoning method which distinguishes the kind of uncertainty is proposed by Matsushima et al. Their research shows clarifying the mathematical model of reasoning method with uncertainty in the statistical standpoint. In this paper, we propose a reasoning system model and a method including uncertainty, based on fuzzy theory. Usually, uncertainty includes not only randomness in field of statistics but also vagueness based on the human's subjective judge with fuzziness. We consider the system in which value of observed fact is calculated using concept of fuzzy events. Moreover, we propose an algorithm of deductive reasoning using Newton-Raphson method, for the reasoning problems with randomness or fuzziness in the system.