An approach of using contexts for in-home activity recognition and forecast

Smart home is one of the most important applications of ubiquitous computing. In this work, we propose an infrastructure of Vietnamese Smart homes as well as a training framework for activity recognition and forecast. In this framework, active learning technique is applied and a new mining algorithm is proposed. In addition to activity recognition, a forecast mechanism is also added into the smart home simulation system by using activity sequence as an extra type of in-home contexts. Experiment results show that the system efficiency is improved when compared to the previous work of Enamul Hoque et al.