A performance comparison of machine learning classification approaches for robust activity of daily living recognition

• Published in: The Artificial intelligence review Vol. 52; no. 1; pp. 357 - 379
• Main Authors: Hussain, Rida Ghafoor, Ghazanfar, Mustansar Ali, Azam, Muhammad Awais, Naeem, Usman, Ur Rehman, Shafiq
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.06.2019; Springer; Springer Nature B.V
• Subjects: Activities of daily living; Activity recognition; Aged; Alzheimer's disease; Artificial Intelligence; Behavior; Change detection; Classification; Comparative analysis; Computer Science; Electronic devices; Employee motivation; Information management; Machine learning; Methods; Older people; Naïve Bayes; Activities of daily living; Bayes Net; K-Nearest Neighbour; Machine learning; Classification; Support Vector Machine
• ISSN: 0269-2821; 1573-7462
• DOI: 10.1007/s10462-018-9623-5

We live in a world surrounded by ubiquitous devices that capture data related to our daily activities. Being able to infer this data not only helps to recognise activities of daily life but can also allow the possibility to recognise any behavioural changes of the person being observed. This paper presents a performance comparison of a series of machine learning classification techniques for activity recognition. An existing hierarchal activity recognition framework has been adapted in order to assess the performance of five machine learning classification techniques. We performed extensive experiments and found that classification approaches significantly outperform traditional activity recognition approaches. The motivation of the work is to enable independent living among the elderly community, namely patients suffering from Alzheimer’s disease.