A study on fuzzy modeling for incorporating human knowledge using fusion axes

Fuzzy modeling is an effective method to construct an interpretable model. This paper proposes a method that incorporates human knowledge using "fusion axes". Human knowledge is often available with simplified expressions. The fusion axes that are proposed by the authors are linear combinations of multiple inputs that are reduce input dimensions for visualizing models with multi-input system. This paper is dealing with a method that makes the simple linguistic expressions by humans and the model in the "fused space" correspond with each other. A simulation is carried out to demonstrate the potentiality of the fusion axes.