Abstract

Activities Recognition (AR) and Occupancy Estimation (OE) are topics of current interest. AR and OE can develop many smart building applications such as energy management and can help
provide good services for residents. Prior research on AR and OE has typically focused on supervised machine learning methods. For a specific smart building domain, a model is trained using data collected from the current environment (domain). The created model will not generalize well when evaluated in a new related domain due to data distribution differences. Creating a model for
each smart building environment is infeasible due to the lack of labeled data. Indeed, data collection is a tedious and time-consuming task. Unsupervised Domain Adaptation (UDA) is a good solution for the considered case. UDA solves the problem of the lack of labeled data in the target domain by allowing knowledge transfer across domains. In this research, we provide several UDA methods that mitigate the data distribution shift between source and target domains using unlabeled target data for OE and AR with and without direct access to labeled source data. Firstly, we consider
techniques that use only a trained source model instead of a huge amount of labeled source data to make domain adaptation. We adapted and tested several UDA methods such as Source HypOthesis Transfer (SHOT), Higher-Order Moment Matching (HoMM), and Source data Free Domain Adaptation (SFDA) on smart building data. Secondly, we adapt and develop several UDA methods that use labeled source data to estimate the number of occupants and recognize activities. The developed methods that have direct access to the source data are the Virtual Adversarial Domain Adaptation (VADA), Sliced Wasserstein Discrepancy (SWD), and Adaptive Feature Norm (AFN). Finally, we make a comparative analysis between several newly adapted deep UDA methods, applied to the tasks of AR and OE, with and without access to labeled source data.