Nonlinear mapping of physical features onto KANSEI space utilizing neural network

摘要

In this paper, we discuss the relationship between physical features and KANSEI features of facial expression. A "EANSEI" space is constructed by using the distance mapping neural network so that the nonlinear mapping of the physical feature into the semantic space is obtianed. From the experimental results, the proposed method is proved to be effective to visualize relationship of data according to the human impressions. Using the proposed method, it will be also possible to construct an individual semantic space for each subject.