The isolation of healthcare data among worldwide hospitals and institutes forms barriers for fully realizing the data-hungry artificial intelligence (AI) models promises in renewing medical services. To overcome this, privacy-preserving distributed learning frameworks, represented by swarm learning and federated learning, have been investigated recently with the sensitive healthcare data retaining in its local premises. However, existing frameworks use a one-size-fits-all mode that tunes one model for all healthcare situations, which could hardly fit the usually diverse disease prediction in practice. This work introduces the idea of ensemble learning into privacy-preserving distributed learning and presents the En-split framework, where the predictions of multiple expert models with specialized diagnostic capabilities are jointly explored. Considering the exacerbation of communication and computation burdens with multiple models during learning, model split is used to partition targeted models into two parts, with hospitals focusing on building the feature-enriched shallow layers. Meanwhile, dedicated noises are implemented to the edge layers for differential privacy protection. Experiments on two public datasets demonstrate En-split’s superior performance on accuracy and efficiency, compared with existing distributed learning frameworks.

Mobile crowd sensing is an innovative paradigm which leverages the crowd, i.e., a large group of people with their mobile devices, to sense various information in the physical world. With the help of sensed information, many tasks can be fulfilled in an efficient manner, such as environment monitoring, traffic prediction, and indoor localization. Task and participant matching is an important issue in mobile crowd sensing, because it determines the quality and efficiency of a mobile crowd sensing task. Hence, numerous matching strategies have been proposed in recent research work. This survey aims to provide an up-to-date view on this topic. We propose a research framework for the matching problem in this paper, including participant model, task model, and solution design. The participant model is made up of three kinds of participant characters, i.e., attributes, requirements, and supplements. The task models are separated according to application backgrounds and objective functions. Offline and online solutions in recent literatures are both discussed. Some open issues are introduced, including matching strategy for heterogeneous tasks, context-aware matching, online strategy, and leveraging historical data to finish new tasks.