Abstract:
Abnormal human behaviors can be signs of a health issue or the occurrence of a hazardous incident. Detecting such behaviors is essential in Ambient Intelligent (AmI) systems to enhance the safety of people. While detecting abnormalities has been extensively explored in different domains, there are still some challenges for developing efficient approaches dealing with the limitations of data-driven approaches to detect abnormal human behaviors in AmI systems. In this paper, a novel approach is proposed to detect such behaviors exploiting the contextual information of human behaviors. Machine-learning models are firstly used to recognize human activities, locations, and objects. Different contexts of human behaviors are then extracted in terms of the duration, frequency, time of the day, locations, used objects, and sequences of the frequent recognized activities. An ontology, called Human ACtivity ONtology (HACON), is proposed to conceptualize the contexts of human behaviors. Finally, a probabilistic version of ASP, a high-level expressive logic-based formalism, is proposed to detect abnormal behaviors through a set of rules based on the HACON ontology. The proposed approach is evaluated in terms of precision, recall, F-measure, and accuracy using two datasets, namely Orange4Home dataset and HAR dataset using smartphones. The evaluation results demonstrate the ability of the proposed approach to detect abnormal human behaviors.
