USER-INDEPENDENT ACTIVITY RECOGNITION METHOD USING GENETIC ALGORITHM BASED FEATURE SELECTION

摘要

A user-independent activity recognition method according to the present invention comprises (a) a step of carrying a mobile terminal with an accelerometer for a plurality of target users, performing a predetermined number of behaviors, and collecting data; (B) a step of extracting a plurality of characteristics for each behavior for each of the target users from the collected data; (C) a step of selecting some features important to the plurality of behaviors and the polarity of the target users; and (d) a step of classifying the user′s behavior using the selected some features. So, the most appropriate features of accelerometer data can be extracted by analyzing vast quantities of features based on the target user and activity behavior.;COPYRIGHT KIPO 2017