FedHAR: Semi-Supervised Online Learning for Personalized Federated Human Activity Recognition

The advancement of smartphone sensors and wearable devices has enabled a new paradigm for smart human activity recognition (HAR), which has a broad range of applications in healthcare and smart cities. However, there are four challenges, privacy preservation, label scarcity, real-timing, and heterogeneity patterns, to be addressed before HAR can be more applicable in real-world scenarios. To this end, in this paper, we propose a personalized federated HAR framework, named FedHAR, to overcome all the above obstacles. Specially, as federated learning, FedHAR performs distributed learning, which allows training data to be kept local to protect users' privacy. Also, for each client without activity labels, in FedHAR, we design an algorithm to compute unsupervised gradients under the consistency training proposition and an unsupervised gradient aggregation strategy is developed for overcoming the concept drift and convergence instability issues in online federated learning process. Finally, extensive experiments are conducted using two diverse real-world HAR datasets to show the advantages of FedHAR over state-of-the-art methods. In addition, when fine-tuning each unlabeled client, personalized FedHAR can achieve additional 10% improvement across all metrics on average.