Fuzzy Reasoning Model consisting of Small Number of Input Rule Modules

Abstract

In the case of the conventional fuzzy reasoning model, the number of the rules increases exponentially with the increase of the input variable. Therefore, it needs too much learning time. This paper proposes SNIRMs (Small Number of Input Rule Modules) conected fuzzy reasoning model. In addition, SNIRMs with deletion of the unnecessary rule modules and deletion of a rule by oblivion learning are proposed to reduce the number of rule modules. In order to show the effectiveness, numerical simulations are performed.