Communication Efficient Federated Learning for Generalized Linear Bandits

Neural Information Processing Systems 

While most existing bandit solutions are designed under a centralized setting (i.e., data is readily available at a central server), in response to the increasing application