Gradient Masked Federated Optimization

Abstract

Federated Averaging (FedAVG) has become the most popular federated learning algorithm due to its simplicity and low communication overhead. We use simple examples to show that FedAVG has the tendency to sew together the optima across the participating clients. These sewed optima exhibit poor generalization when used on a new client with new data distribution. Inspired by the invariance principles in (Arjovsky et al., 2019; Parascandolo et al., 2020), we focus on learning a model that is locally optimal across the different clients simultaneously. We propose a modification to FedAVG algorithm to include masked gradients (AND-mask from (Parascandolo et al., 2020)) across the clients and uses them to carry out an additional server model update. We show that this algorithm achieves better accuracy (out-of-distribution) than FedAVG, especially when the data is non-identically distributed across clients.

References in corpus (7)

• Federated Learning: Strategies for Improving Communication Efficiency
• Federated Optimization in Heterogeneous Networks
• Invariant Risk Minimization
• On the Convergence of Local Descent Methods in Federated Learning
• Learning explanations that are hard to vary
• Federated Learning with Unbiased Gradient Aggregation and Controllable Meta Updating
• Predicting with High Correlation Features