A Study for Statistical Machine Learning to Classify Radial Motions of the Elderly

Abstract

Human visual information in peripheral and foveal vision fields is processed along dorsal and ventral streams in the brain, respectively. It has been also reported that the motion sickness is induced by visual approaches especially in the peripheral vision. In general, difference in the radial motion has been found between the peripheral and the foveal vision. In this study, we measured eye movements during the peripheral or foveal vision of 3D video clips and conducted statistical machine learning to classify sample data into two types of radial motion processes. We measured the eye movements of six healthy elderly people while viewing 3D video clips. We extracted about 2,300 time series data of these eye movements for 1, 2, 3, 4, 5, 10, and 20 seconds, respectively. These time series were used as a data set, and statistical machine learning was performed for the classification. As a result, the accuracy of the classification increased with the sample size. In addition, a comparison of the correct rate of the classification between peripheral and foveal vision showed that the rate was higher for the foveal vision.