Entropy measures for anomaly detection

Howedi, A.F.O., 2020. Entropy measures for anomaly detection. PhD, Nottingham Trent University.

[img]
Preview
Text
Aadel Howedi 2020.pdf - Published version

Download (6MB) | Preview

Abstract

Human activity recognition methods are used to support older adults to live independently in their own homes by monitoring their Activities of Daily Living (ADL). The gathered data and information representing different activities will be used to identify anomalous activities in comparison with the routine activities. In the related research in this area, the most recent studies have mainly focused on detecting anomalies in a single occupant environment. Although older adults often receive visits from family members or health care workers, representing a multi-occupancy environment.

This research is focused on the application of entropy measures for anomaly detection in ADLs in a single-occupancy and multi-occupancy environment. In many applications, entropy measures are used to detect the irregularities and the degree of randomness in data. However, this has rarely been applied in the context of activities of daily living.

To address the research questions identified in the thesis, three novel contributions of the thesis are; Firstly, a novel method based on different entropy measures is investigated to detect anomalies in ADLs, specifically in sleeping routine and human falls. Secondly, a novel entropy-based method is explored to detect anomalies in ADLs in the presence of a visitor, solely based on information gathered from ambient sensors. Finally, entropy measures are applied to investigate their effectiveness in identifying a visitor in a single home environment based on data gathered from ambient sensors. The results presented in this thesis show that entropy measures could be used to detect abnormality (here, irregular sleep, human fall and a visitor) in ADLs.

Item Type: Thesis
Creators: Howedi, A.F.O.
Date: December 2020
Divisions: Schools > School of Science and Technology
Record created by: Linda Sullivan
Date Added: 26 Nov 2021 11:52
Last Modified: 26 Nov 2021 11:52
URI: https://irep.ntu.ac.uk/id/eprint/44958

Actions (login required)

Edit View Edit View

Views

Views per month over past year

Downloads

Downloads per month over past year