ベクトル量子化を用いたファジィルールの学習法

抄録

We propose a learning method of fuzzy inference rules using vector quantization, neural gas network. Some models using self-organization or vector quantization by neural networks are proposed in recent stud-ies. These models show good results as for the number of fuzzy inference rules in high dimensional problems. However, most of these models determine a distribution of initial fuzzy inference rules by considering only input data. In this paper, so as to make a more proper distribution of the initial fuzzy inference rules in input space, we propose a method considering not only input data but output data. Further, the number of fuzzy inference rules is determined to an objective value(threshold of inference error) by a constructive way. In order to demonstrate the validity of the proposed method, some numerical examples are performed.