1C3-4 関数重み付きSIRMs結合型推論モデル(ソフトコンピューティング)

概要

論文の詳細を見る
This paper discusses the generalization of a SIRMs (Single-Input Rule Modules) model. A SIRMs model consists of a number of groups of simple fuzzy if-then rules with only a single attribute in the antecedent part. The final outputs of conventional SIRMs model are obtained by summarizing product of the weight and inference result from a rule module. In the SIRMs model of the paper, we propose a SIRMs model with functional weights where a linear function is used in the weights. Triangular-type membership functions are used for the antecedent part of fuzzy if-then rules. The gradient decent method is applied to a proposed SIRMs model in order to derive update rules for minimizing the error between the target and output from the model. In the computational experiments of this paper, the proposed SIRMs model is applied to a modeling domain of nonlinear functions.
2011-09-01