Abstract

In many large-scale machine learning applications, data is acquired and processed at the edge nodes of the network such as mobile devices, users’ devices, and IoT sensors. While distributed learning at the edge can enable a variety of new applications, it faces major systems bottlenecks that severely limit its reliability and scalability including system and data heterogeneity and communication bottleneck. In this talk, we focus on federated learning which is a new distributed machine learning approach, where a model is trained over a set of devices such as mobile phones, while keeping data localized. We first present a straggler-resilient federated learning scheme that uses adaptive node participation to tackle the challenge of system heterogeneity. We next present a robust optimization formulation for federated learning that enables us to address the data heterogeneity challenge in federated learning. We finally talk about a new algorithm for personalizing the learned models for different users.