A New Approach to Recognize Activities in Smart Environments Based on Cooperative Game Theory

These days, a lot number of elderly people need health care which may cause huge financial costs, especially in formal case. Machine Learning and the profound achievements in sensing technology provide the opportunities to monitor people living independently at home and can detect a distress situation affordably. Although there are some approaches to do recognize activities for this purpose, but there has not been any game-theoretic approach in order to select the most efficient sensors to reduce the system's overhead by decreasing the number of features. In this paper, we present a new classifier to recognize activities in a smart environment that is based on selection of most efficient sensors by cooperative game theory. The sensors are selected in which provide more information about the target classes. We show the performance of our algorithm by simulation.