Classification and Visualization of Long-Term Life-monitoring Sensor Signals Using Topological Characteristics of Category Maps

摘要

This paper presents a novel extraction and visualization method of human behavior patterns as life rhythms from sensor signals obtained using our originally developed life-monitoring system. Our method visualizes categorical relations and distribution characteristics on category maps and their fired units. For creating category maps that preserve data topology, we optimized three main parameters: vigilance thresholds, mapping size, and learning iterations. The mapping size related to classification granularity and expression ability must be changed along with analysis of the data length. Experimentally obtained results reveal that the distribution of burst units is spread evenly along with the setting of learning iterations greater than the data size. This characteristic indicates that it is necessary to increase learning iterations when the mapping size is increased. Moreover, we demonstrate characteristics of integration and division of categories for the relation between fired units and category maps with changing of the vigilance parameter.