Recognizing Humans from Their Behavioral Patterns
Sonia Sonia · 2019
Authentication of human in an unobtrusive manner is important in modern technology to enrich society. Authentication of a person in an environment like smart home, car driving, etc., can lead to better man-machine conjugation. Advancement in sensor technology and machine learning makes it possible to obtain and analyze a significant amount of data. This paper explores the important aspects of human behavior daily routine in the home. Traditional approaches rely on the extraction of statistical features for machine learning algorithms. However, with an increased number of sensors, feature extraction may not be possible. Therefore, human identification in a smart home is performed using deep learning technique on the data obtained from multiple sensors placed at the various place and devices in the homes like a bed, bathroom, door. In this paper, both statistical, as well as deep neural networks, are implemented for human activity data. For humans in smart home deep learning has upper hand as compared to other machine learning techniques. Thus proposed approach can identify humans based on analysis of sensor data using various machine learning technique using unobtrusive sensory data.