REAL-TIME DETECTION OF MULTI-SCALE CHANGES IN SMART HOMES

Abstract

Change Point Detection (CPD) is the problem of discovering time points at which the behavior of a time series changes abruptly. Such abrupt changes may represent transitions that occur between states. Tracking these changes from smart home sensor data can help older adults and their caregivers to manage their daily lives, monitor their health condition, and receive alerts regarding emergency situations. In this work, we present a novel near-real time nonparametric change point detection algorithm called SEP, which uses Separation distance as a divergence measure to detect change points in high-dimensional time series. Through experiments on artificial and real-world datasets, we demonstrate the usefulness of the proposed method in comparison with existing methods for these types or problems. We also use this algorithm to detect activity transitions in smart homes. Real time detection of transitions between activities based on sensor data is a valuable but somewhat untapped challenge. Detecting these transitions is useful for activity segmentation, for timing notifications or interventions, and for analyzing human behavior.