Collaborative learning for control

We begin by studying the problem of learning a linear system model from the observations of M clients. The catch: Each client is observing data from a different dynamical system. This work addresses the question of how multiple clients collaboratively learn dynamical models in the presence of system and environmental heterogeneity. We pose this problem as a federated learning problem and characterize the tension between achievable performance and system heterogeneity. Our federated sample complexity result provides a constant factor improvement over the single agent setting in the low heterogeneity regime. We then show how clustering can improve these results. Finally, we will extend this framework to the setting of federated optimal control design using policy gradients.