Federated learning (FL) has emerged as a popular machine learning paradigm which allows multiple data owners to train models collaboratively without sharing their raw datasets. It holds potential for a wide spectrum of analytics applications on sensitive data.Federated learning (FL) has emerged as a popular machine learning paradigm which allows multiple data owners to train models collaboratively without sharing their raw datasets. It holds potential for a wide spectrum of analytics applications on sensitive data.[#item_full_content]