Acquisition of KANSEI based on Fuzzy inference and multivaritate analysis

摘要

Human brain mainly performs two kinds of information processes, intelligent process and KANSEI process. KANSEI is used for the sensory test or evaluation in various fields. In this paper, we propose an expression method of KANSEI. This proposed method is shown by the following steps. First, we prepare multiple characteristics, which are standard indexes for evaluating the object. Next, we examine the correlation between characteristics and evaluation value of the examinee. Then we estimate significant characteristics for the examinee by using the multivariate analysis. Finally, we express KANSEI of the examinee by fuzzy reasoning using spline function on limited characteristic. In this method, it is possible to determine the parameter that affects the evaluation, and it is proven characteristics that the examinee has the interest. This proposed method uses a self-tuning fuzzy inference based on the spline membership function to express individual KANSEI. It is also advantageous to make the forecasting evaluation about the information that the expert yet does not observe, by continuously expressing membership function using the spline function. To show the effectiveness of the proposed system, we applied this method to evaluation of persimmons and eggplants. It is considered that there is mainly an individual difference about the color of the persimmon, and the form of eggplant. This method could be applied to various kinds of sensory evaluating systems.