A Survey of User-Centred Approaches for Smart Home Transfer Learning and New User Home Automation Adaptation

Abstract

Recent smart home applications enhance the quality of people’s home experiences by detecting their daily activities and providing them services that make their daily life more comfortable and safe. Human activity recognition is one of the fundamental tasks that a smart home should accomplish. However, there are still several challenges for such recognition in smart homes, with the target home adaptation process being one of the most critical, since new home environments do not have sufficient data to initiate the necessary activity recognition process. The transfer learning approach is considered the solution to this challenge, due to its ability to improve the adaptation process. This paper endeavours to provide a concrete review of user-centred smart homes along with the recent advancements in transfer learning for activity recognition. Furthermore, the paper proposes an integrated, personalised system that is able to create a dataset for target homes using both survey and transfer learning approaches, providing a personalised dataset based on user preferences and feedback.